Cation Exchange as a Route to Introduce Magnetism to Hybrid-Improper Polar Phases

Abstract

The pseudo Ruddlesden–Popper phase Li₂CaTa₂O₇ is converted to ZnCaTa₂O₇, FeCaTa₂O₇, or CoCaTa₂O₇ by reaction with the corresponding transition-metal dichloride. Diffraction data reveal that ZnCaTa₂O₇ adopts a polar crystal structure (P2cm) with the Zn²⁺cations ordered into stripes within the interlayer coordination sites, and the TaO₆ units adopt an a ^– b ^– c ⁺/–(a ^– b ^–)c ⁺ tilting pattern. In contrast, FeCaTa₂O₇ and CoCaTa₂O₇ adopt polar structures (P2₁ nm) with the transition-metal cations ordered in a checkerboard pattern within the interlayer coordination sites, and the TaO₆ units adopt an a ^– b ^– c ⁺/ b ^– a ^– c ⁺ tilting pattern. The different polar structures adopted are rationalized on the basis of the size of the interlayer transition-metal cation. On cooling, FeCaTa₂O₇ (T _N = 40 K) and CoCaTa₂O₇ (T _N = 25 K) adopt antiferromagnetically ordered states with spins aligned parallel to the crystallographic stacking axis and arranged in a G-type manner. Close inspection of the NPD data collected from FeCaTa₂O₇ at low temperature reveals a diffuse component to the magnetic scattering, which, in combination with magnetization data, suggest a glassy component to the low-temperature magnetic state. Neither FeCaTa₂O₇ nor CoCaTa₂O₇ shows significant lattice parameter anomalies around their respective Néel temperatures, in contrast to the previously reported manganese analogue MnCaTa₂O₇.

Introduction

Materials which simultaneously exhibit a spontaneous, switchable electrical polarization (ferroelectricity) and a spontaneous, switchable magnetic polarization (ferromagnetism) are highly desired because such magnetoelectric materials − could allow the preparation of novel devices with applications in data manipulation and storage. , However, the preparation of novel magnetoelectric materials is challenging because the two constituent behaviors, ferromagnetism and ferroelectricity, are contraindicated. This is principally because the noncentrosymmetric structural distortions which are a prerequisite of ferroelectric behavior are typically driven by a second-order Jahn–Teller (SOJT) instability arising from the presence of either octahedrally coordinated d⁰ transition-metal cations (e.g., Ti⁴⁺ in BaTiO₃) , or ns² post-transition-metal cations (e.g., Pb²⁺ in PbZrO₃) , and these species are diamagnetic, so cannot contribute to the desired ferromagnetic behavior.

Recently an alternative ‘trilinear-coupled hybrid-improper’ mechanism for stabilizing ferroelectric structural distortions has been attracting attention. , In this scheme, two nonpolar structural distortions (typically the tilting distortions of layered perovskite phases) couple together to stabilize a third, polar distortion (trilinear coupling) which is energetically unfavorable in the absence of the nonpolar distortions. The polar distortion stabilized by this mechanism is not the primary order parameter of the associated ferroelectric phase transition, so polar materials of this type are often referred to as ‘hybrid improper’ ferroelectrics. ,

In principle the hybrid-improper stabilization mechanism does not impose any restrictions on the chemical composition of polar phases, as it is geometric in nature. Thus, paramagnetic species should be easy to incorporate into hybrid-improper ferroelectric phases in an attempt to prepare magnetoelectric materials, as demonstrated by Ca₃Mn₃O₇, the first material theoretically identified as a hybrid-improper ferroelectric. In practice, however, the requirement to have two distinct octahedral tilting distortions does impose quite strict chemical constraints on the makeup of hybrid improper ferroelectric phases because such highly distorted frameworks are only stable in materials with Goldschmitt tolerance factors (t = < A-O>/√2 < B–O>) smaller than around t = 0.87. , This small tolerance factor excludes a large number of combinations of A- and B-cations, including the majority of Ruddlesden–Popper and Dion-Jacobson phases which can be prepared with paramagnetic transition-metal cations on their B-sites.

Recently we have been using the facile exchange reactions of the monovalent cations in A’AB ₂O₇ (A’ = Cs, Rb) Dion-Jacobson − and Li₂ AB ₂O₇ pseudo Ruddlesden–Popper phases to prepare metastable layered perovskite oxides with tolerance factors which are small enough to exhibit polar structures stabilized by the hybrid-improper mechanism. Building on this we have also shown that substitution of 2 Li⁺ cations with a divalent paramagnetic ion, such as Mn²⁺, allows magnetic behavior to be introduced. , Thus, for example, the hybrid-improper polar phase Li₂SrTa₂O₇ can be converted to MnSrTa₂O₇, a paramagnetic polar material which exhibits signatures of magnetoelectric coupling.

Here we describe the preparation and characterization of FeCaTa₂O₇, CoCaTa₂O₇ and ZnCaTa₂O₇ prepared by analogous cation exchange reactions from Li₂CaTa₂O₇, like the previously reported phase MnCaTa₂O₇.

Experimental Section

Synthesis of Li₂CaTa₂O₇

Polycrystalline samples of Li₂CaTa₂O₇ were prepared by combining suitable stoichiometric ratios of CaCO₃ (99.997%) and Ta₂O₅ (99%, dried at 900 °C) with a 3% stoichiometric excess of Li₂CO₃ (99.99%) to compensate for lithium loss due to volatility at high temperature. The mixture was ground in an agate pestle and mortar, placed in an alumina crucible and heated at 800 °C in air for 12 h. The mixture was then reground, pressed into 13 mm pellets, placed in an alumina crucible and heated to 1200 °C at a rate of 5 °C/min and then held there for 2 h. The sample was then removed from the furnace, reground and pressed into pellets and then directly inserted into a furnace at 1200 °C and heated for 2 h, before being quenched to room temperature. Synchrotron X-ray diffraction data collected from Li₂CaTa₂O₇ could be fit well by the reported structure , to yield lattice parameters in good agreement with literature values, as described previously.

Synthesis of FeCaTa₂O_7, CoCaTa₂O₇ and ZnCaTa₂O₇

Polycrystalline samples of FeCaTa₂O_7, CoCaTa₂O₇ and ZnCaTa₂O₇ were prepared via cation exchange reactions from Li₂CaTa₂O₇. Approximately 4 g of Li₂CaTa₂O₇ was combined with 10-mol equiv of anhydrous FeCl₂ (99.5%), CoCl₂ (99.7%) or ZnCl₂ (99.99%). The mixtures were ground in an agate pestle and mortar in an argon-filled glovebox, loaded into Pyrex tubes and sealed under vacuum. The separate samples of FeCaTa₂O_7, CoCaTa₂O₇ and ZnCaTa₂O₇ were then heated at a ramp rate of 2 °C/min to 350, 360, and 400 °C, respectively, for two periods of 48 h, followed by cooling at 5 °C/min to room temperature. The reaction tube containing ZnCl₂ was placed vertically in a furnace because ZnCl₂ has a melting point of 290 °C so the reaction medium is molten at the synthesis temperature. Between heating cycles, the samples were washed with distilled water to remove any unreacted transition-metal chloride, and LiCl byproduct, and then combined with fresh anhydrous transition-metal chloride for the following cycle.

Characterization

X-ray powder diffraction data were collected using a PANalytical X’pert diffractometer incorporating an X’celerator position-sensitive detector (monochromatic Cu Kα₁ radiation). High-resolution synchrotron X-ray powder diffraction (SXRD) data were collected using the I11 instrument at the Diamond Light Source Ltd. Diffraction patterns were collected using Si-calibrated X-rays with an approximate wavelength of 0.825 Å from samples, sealed in 0.3 mm diameter borosilicate glass capillaries. Time-of-flight neutron powder diffraction (NPD) data were collected using the WISH diffractometer located at the ISIS neutron source, from the samples loaded in vanadium cans. Rietveld refinements were performed using the TOPAS Academic (V6). Second harmonic generation (SHG) response of samples was measured from powder samples with the SHG intensity compared to a standard samples of potassium dihydrogen phosphate (KDP) or AgGaS₂. No index matching fluid was used in any of the experiments. A detailed description of the experimental setup and process has been reported previously. DC magnetization data were collected using a Quantum Design MPMS SQUID magnetometer from samples contained in gelatin capsules.

Results

Structural Characterization of ZnCaTa₂O₇

Powder SHG data collected from ZnCaTa₂O₇, using a laser of wavelength 1064 nm, exhibit an activity that is 0.07 times that of a KDP standard, indicating that ZnCaTa₂O₇ adopts a noncentrosymmetric structure (Figure S1). SXRD data collected from ZnCaTa₂O₇ at room temperature can be indexed using a primitive orthorhombic unit cell (a = 5.382 Å, b = 5.538 Å, c = 19.764 Å) consistent with a √2 × √2 × 1 geometric expansion of an undistorted n = 2 Ruddlesden–Popper unit cell, and the data showed no indications of any secondary phases.

NPD data collected from ZnCaTa₂O₇ at 100 K could also be indexed using a primitive orthorhombic unit cell (a = 5.38116(7) Å, b = 5.54170(7) Å, c = 19.78286(25) Å). A detailed symmetry analysis of the n = 2 Ruddlesden–Popper structure was used to generate a series of distorted, chemically plausible structural models for ZnCaTa₂O₇ based on the cooperative tilting of the TaO₆ units, as described previously. By considering the reflection conditions observed in the NPD data, and the fact that ZnCaTa₂O₇ adopts a noncentrosymmetric crystal structure, it was possible to narrow down the list of possible distorted structures to two candidates: an a ^– b ^– c ⁺/b ^– a ^– c ⁺ distorted structure described in space group P2₁ nm (#31) or an a ^– b ^– c ⁺/-(a ^– b ^–)c ⁺ distorted structure described in space group P2cm (#28). Thus, structural models were constructed for the two distortions based on the structure of Li₂CaTa₂O₇ but with each lithium replaced by 0.5 Zn²⁺ ions. Refinement of these Zn-disordered models against the NPD data revealed the P2cm symmetry model gave a better fit to the data (P2cm: wRp = 7.47; P2₁ nm: wRp = 13.07).

The possibility of Zn occupational-order was then considered because the two structural models have multiple crystallographically distinct Zn sites: 4 distinct sites in the P2cm model (2 × 2a, 2 × 2b) and 2 distinct sites in the P2₁ nm model (2 × 4b). Ordering the Zn cations onto one of the 4b sites within the P2₁ nm model would lead to a chequerboard ordering of the Zn cations within the ZnO layers. However, refinement of the Zn site occupancies within the P2₁ nm model led to no variation of the disordered distribution of the Zn cations and no improvement to the fit to the data (wRp = 13.07). In contrast, refinement of the Zn site occupancies within the P2cm model (within the constraint that the ZnCaTa₂O₇ composition was maintained) rapidly lead to an occupancy of 0.98(3) for one 2a and one 2b site with the remaining sites having occupancies of 0.02(3) yielding a structure in which the Zn cations are arranged in stripes within the ZnO layers. This Zn-ordered model fit the NPD data better than the Zn-disorder model (wRp = 4.76) and setting the occupancies of the filled sites to unity made no difference to the fitting statistics. We therefore conclude that the structure of ZnCaTa₂O₇ is best described by a Zn-stripe ordered model with an a ^– b ^– c ⁺/-(a ^– b ^–)c ⁺ tilting distortion described in space group P2cm as detailed in Table S1, with selected bond lengths given in Table S2, fits to the data shown in Figure S2 and a representation of the structure shown in Figure .

1.

Crystal structures of ZnCaTa₂O₇ and FeCaTa₂O₇ viewed down their respective [110] axes. Insets show interlayer planes viewed down respective [001] axes.

Structural Characterization of FeCaTa₂O₇

Powder SHG data collected from FeCaTa₂O₇, using a laser of wavelength 2090 nm, indicate an activity that is 0.4 times that of an AgGaS₂ standard, indicating that FeCaTa₂O₇ adopts a noncentrosymmetric structure (Figure S1).

SXRD data collected from FeCaTa₂O₇ at room temperature could be indexed using a metrically tetragonal unit cell (a = 5.513 Å, c = 18.658 Å) as could the NPD data collected at 200 K. However, symmetry analysis of the n = 2 Ruddlesden–Popper system revealed that there are no cooperative tilting distortions which yield tetragonal, noncentrosymmetric structures, so collective distortions which yield orthorhombic noncentrosymmetric structures were considered. Of the 4 tilting distortions which yield noncentrosymmetric orthorhombic structures (a ^– a ^– c ⁺/a ^– a ^– c ⁺, A2₁ am; a ^– b ^– c ⁺/ b ^– a ^– c ⁺, P2₁ nm; a ^– a ^– c ⁺/-(a ^– a ^– c ⁺), B2cm; a ^– b ^– c ⁺/-(a ^– b ^–)c ⁺, P2cm) only those described in space groups P2₁ nm and P2cm have reflection conditions compatible with the NPD data. Thus, structural models were constructed for FeCaTa₂O₇ in the space groups P2₁ nm and P2cm based on the structure of Li₂CaTa₂O₇ but with each lithium ion replaced by 0.5 Fe²⁺ cations.

Refinement of these models against the NPD data revealed the P2₁ nm symmetry model gave a much better visual and statistical fit (wRp = 7.39) to the data than the P2cm model (wRp = 12.13), so the a ^– b ^– c ⁺/ b ^– a ^– c ⁺ distorted, P2₁ nm symmetry model was chosen to describe the structure of FeCaTa₂O₇. Close inspection of the P2₁ nm model revealed the Fe²⁺ cations can reside on 2 separate 4b crystallographic sites, allowing Fe occupationally ordered structures to be described. Refinement of the occupancies of these two sites, under the constraint that the FeCaTa₂O₇ composition was conserved, led to a rapid filling of one site and emptying of the other to yield a fully Fe-cation ordered structure. This was accompanied by a significant improvement to the visual and statistical fit (wRp = 4.25). Full details of the refined structure of FeCaTa₂O₇ are described in Table S3, with selected bond lengths given in Table S4, fits to the data shown in Figure S3 and a representation of the structure shown in Figure .

It should be noted that a cooperative a ^– b ⁰ c ⁰/ b ^– a ⁰ c ⁰ distortion of an n = 2 framework yields a tetragonal structure described in space group P4₂/mnm, which can be seen as a special case of the a ^– b ^– c ⁺/ b ^– a ^– c ⁺ distortion refined for FeCaTa₂O₇. Close inspection of the refined structure of FeCaTa₂O₇ reveals that while the a-tilt angle (20.3(4)°) is significantly larger than the b-tilt (2.1(4)°) or c-tilt (5.7(8)°) angles, the latter two tilts have nonzero magnitudes, confirming the choice of a noncentrosymmetric orthorhombic P2₁ nm symmetry model, rather than a centrosymmetric tetragonal P4₂/mnm model.

Structural Characterization of CoCaTa₂O₇

Powder SHG data collected from CoCaTa₂O₇, using a laser of wavelength 2090 nm, indicate an activity that is 0.7 times that of an AgGaS₂ standard, indicating that CoCaTa₂O₇ adopts a noncentrosymmetric structure (Figure S1). SXRD data collected from CoCaTa₂O₇ at room temperature could be indexed using a metrically tetragonal unit cell (a = 5.516 Å, c = 18.59 Å) as could the NPD data collected at 100 K. Using logic analogous to that described above for the structural analysis of FeCaTa₂O₇, a ^– b ^– c ⁺/ b ^– a ^– c ⁺ and a ^– b ^– c ⁺/-(a ^– b ^–)c ⁺ distorted structural models were constructed for CoCaTa₂O₇ in the space groups P2₁ nm and P2cm respectively, based on the structure of Li₂CaTa₂O₇ but with each lithium ion replaced by 0.5 Co²⁺ cations.

As cobalt has a relatively weak neutron scattering power (2.49 fm) these structural models were simultaneously refined against both the NPD and SXRD data collected at 100 K. Again, the P2₁ nm model gave a better visual and statistical fit (wRp = 4.78) to the data than the P2cm model (wRp = 6.45), and when the occupancies of the cobalt sites were refined a fully Co occupationally ordered model resulted, again with a significant improvement to the visual and statistical fit (wRp = 2.07). Full details of the refined structure of CoCaTa₂O₇ are described in Table S5, with selected bond lengths given in Table S6, fits to the data shown in Figure S4 and S5.

Close inspection of the refined structure of CoCaTa₂O₇ revealed the a-tilt angle (23.0(3)°) was again significantly larger than the b-tilt (1.1(3)°) or c-tilt (1.7(7)°) angles. However, all three tilt angles are have nonzero magnitudes, confirming the choice of a noncentrosymmetric orthorhombic P2₁ nm symmetry model, rather than a centrosymmetric tetragonal P4₂/mnm model for CoCaTa₂O₇.

Magnetic Characterization of FeCaTa₂O₇

Magnetization data collected from FeCaTa₂O₇ in an applied field of 100 Oe (Figure ) can be fit by the Curie–Weiss law in the temperature range 150 < T/K < 300, yielding values of C = 4.21(3) cm³ K mol^–1 and θ = −107.2(5) K. The observed Curie constant is significantly larger than that expected for a spin-only S = 2 ion (C _expected = 3 cm³ K mol^–1) suggesting an unquenched orbital contribution to the moment via second-order spin–orbit coupling. The ZFC and FC data diverge weakly below 120 K which we attribute to the presence of a small quantity of Fe₃O₄ (Verwey transition T ∼ 120 K). There is a much stronger divergence between ZFC and FC data observed below T ∼ 50 K and a local maximum in the ZFC data at T = 38 K which is accompanied by an inflection in the FC data. Magnetization data collected as a function of applied field at 300 K are linear and pass through the origin (Figure ). Analogous data collected at 5 K, after cooling in an applied field of 5 T, are displaced from the origin suggesting a glassy component to the magnetic behavior, although AC susceptibility data collected in the range 35 < T/K < 45 show no strong frequency dependence as described in detail in the Supporting Information.

2.

(top) ZFC and FC data collected from FeCaTa₂O₇ in an applied field of 100 Oe. Inset shows fit to Curie–Weiss law. (bottom) magnetization-field data collected from FeCaTa₂O₇ at 5 and 300 K.

NPD data collected from FeCaTa₂O₇ at 1.5 K exhibit a series of additional reflections, not observed in analogous data collected at 200 K (Figure a), which are attributed to magnetic order. These additional reflections can be indexed by the crystallographic unit cell, suggesting a propagation vector k = (0, 0, 0). A series of magnetic models were constructed on this basis using the ISODISTORT software package , and refined against the NPD data. The best fit was achieved using a model obtained by applying the mΓ₂ magnetic irreducible to the P2₁ nm crystallographic structure to yield a model described in magnetic space group P2₁ n’m’ (#31.127). The components of the ordered moments parallel to the x- and y-axes refined to zero, within error, while the z-component converged to a value of 3.09(1) μ_B, yielding a model which can be thought of as a G-type antiferromagnetic ordering, as shown in the inset to Figure b and described in detail in Table S7.

3.

a) NPD data collected from FeCaTa₂O₇ at temperature indicated. Arrows indicate magnetic Bragg peaks. b) Plot of ordered magnetic moment as a function of temperature. Inset shows magnetic structure of FeCaTa₂O₇ (blue, gray and green and red spheres represent Ta, Ca, Fe, and O respectively).c) Expanded view of NPD data showing diffuse magnetic scattering at d ∼ 5.3 Å.

NPD data collected from FeCaTa₂O₇ on warming from 1.5 K can be fit by the same combined nuclear and magnetic model and reveal the ordered moment of the system declines with increasing temperature, as shown in Figure a, with no magnetic diffraction intensity observed above 40 K. The temperature dependence of the ordered moment shown in Figure b cannot easily be fit by a power law.

Close inspection of the fit of the combined structural and magnetic model to the NPD data collected from FeCaTa₂O₇ at 1.5 K reveal a broad, weak diffraction feature centered at d ∼ 5.3 Å, as shown in Figure c, which declines in intensity on warming in a manner akin to the main magnetic scattering features. The presence of this diffuse magnetic scattering helps to explain the relatively small size of the observed ordered moment on the Fe centers (3.09 μB compared to an expected value of 4 μB) and when combined with the displaced magnetization-field data shown in Figure , suggests there is a disordered or glassy component to the low temperature magnetic state.

Magnetic Characterization of CoCaTa₂O₇

Magnetization data collected from CoCaTa₂O₇ in an applied field of 100 Oe (Figure ) can be fit by the Curie–Weiss law in the temperature range 100 < T/K < 300, yielding values of C = 1.988(1) cm³ K mol^–1 and θ = −38.4(8) K, and a temperature independent contribution of 1.75(6) × 10^–3 cm³ mol^–1. Again, the observed Curie constant is significantly larger than that expected for a spin-only S = ³/₂ ion (C _expected = 1.875 cm³ K mol^–1) consistent with second-order spin–orbit coupling. On cooling below T = 50 K both ZFC and FC data exhibit a sharp increase before diverging at T = 28 K. Magnetization-field data collected at 300 K (Figure ) are linear and pass through the origin, while data collected at 5 K, after cooling in an applied field of 5 T, are sigmoidal and exhibit hysteresis, consistent with canted antiferromagnetic behavior. Analogous data collected on warming show no hysteresis above 28 K and become linear at T = 35 K as shown in the Supporting Information.

4.

(top) ZFC and FC data collected from CoCaTa₂O₇ in an applied field of 100 Oe. Inset shows fit to Curie–Weiss law after subtraction of the temperature independent component. (bottom) magnetization-field data collected from CoCaTa₂O₇ at 5 and 300 K.

NPD data collected from CoCaTa₂O₇ at 1.5 K exhibit a series of additional reflections not observed in analogous data collected at 100 K (Figure a), which are attributed to magnetic order. In common with FeCaTa₂O₇ the additional reflections can be indexed using the crystallographic cell and are best fit by a model obtained by applying the mΓ₂ magnetic irreducible to the P2₁ nm crystallographic structure to yield a model described in space group P2₁ n’m’ (#31.127). On refinement the components of the ordered moments parallel to the x- and y-axes refined to zero, within error, while the z-component converged to a value of 3.07(1) μ_B, yielding a model directly analogous to that of FeCaTa₂O₇, shown in Figure and described in detail in Table S8.

5.

a) NPD data collected from CoCaTa₂O₇ at temperature indicated. Arrows indicate magnetic Bragg peaks. b) Plot of ordered magnetic moment as a function of temperature.

NPD data collected from CoCaTa₂O₇ on warming from 1.5 K can be fit by the same combined nuclear and magnetic model and reveal the ordered moment of the system declines with increasing temperature, as shown in Figure b, with no magnetic diffraction intensity observed above 25 K. Again, the temperature dependence of the ordered moment cannot easily be fit by a power law.

Discussion

Substitution of the lithium cations in Li₂CaTa₂O₇ with Co²⁺ or Fe²⁺ yields MCaTa₂O₇ phases which are isostructural with MnCaTa₂O₇ (space group P2₁ nm) in which the transition metal cations adopt a chequerboard vacancy-ordered arrangement within the sites previously occupied by lithium (Figure ) which stabilizes an a ^– b ^– c ⁺/ b ^– a ^– c ⁺ tilting distortion of the TaO₆ octahedra. In contrast, the Zn substituted phase ZnCaTa₂O₇ adopts a structure in which the Zn²⁺ cations are ordered into stripes within the interlayer coordination sites (Figure ) which stabilizes an a ^– b ^– c ⁺/-(a ^– b ^–)c ⁺ tilting distortion of the TaO₆ octahedra, described in space group P2cm.

Thus, we can see that both the P2₁ nm symmetry structure adopted by the Mn, Co and Fe phases, and the P2cm symmetry structure adopted by ZnCaTa₂O₇ contain a ^– b ^– c ⁺ tilted CaTa₂O₇ perovskite layers, with one difference between the two structure types being how the distortions in adjacent layers are oriented relative to each other. In the P2₁ nm symmetry, a ^– b ^– c ⁺/ b ^– a ^– c ⁺ distorted structure there is a 90° rotation around the z-axis of the tilt configuration between adjacent a ^– b ^– c ⁺ distorted layers. In the P2cm symmetry a ^– b ^– c ⁺/-(a ^– b ^–)c ⁺ distorted structure there is an inversion in the direction of the out-of-phase tilts in the xy-plane between adjacent a ^– b ^– c ⁺ distorted layers.

This difference between the adjacent-layer orientation of the CaTa₂O₇ tilting distortions can be attributed to the differing interlayer cation ordering schemes of the phases: chequerboard order for MCaTa₂O₇ (M = Mn, Fe, Co), stripe order for ZnCaTa₂O₇. Analogous interlayer cation ordering patterns are observed in LiNdNb₂O₇ (stripes) and NaNdNb₂O₇ (chequerboard) – two phases prepared via cation exchange from RbNdNb₂O₇. In these ANdNb₂O₇ phases the differing cation ordering arrangements adopted by the Li and Na phases are rationalized by considering the competition between the desire to minimize A-A cation repulsion, and the need to optimize the metal–oxygen bonding in the AO₄ local coordination polyhedra. The chequerboard cation ordering in NaNdNb₂O₇ consists of sheets of corner-linked NaO₄ units, and thus minimizes Na–Na cation repulsion in line with Pauling’s third crystallographic rule. However, this cation arrangement does not allow the size of the NaO₄ coordination polyhedra to be modified via the tilting of the NbO₆ octahedra. The stripe-ordered structure of LiNdNb₂O₇ consists of sheets of edge-sharing LiO₄ units and thus has a higher degree of A-A cation repulsion than the chequerboard arrangement, but crucially in this configuration the tilting of the NbO₆ units can optimize the size of the LiO₄ polyhedra. Thus, large cations (i.e., Na⁺) adopt a chequerboard ordered structure to minimize A-A repulsion, while small cations (i.e., Li⁺) adopt stripe ordered structures because optimizing (shortening) the A-O bonds in the local AO₄ units becomes the energetic priority.

This rationalization can be transferred directly to the MCaTa₂O₇ phases, with the larger M ²⁺ cations (Mn, Fe, Co) adopting chequerboard ordered structures and the smaller M ²⁺ cations (Zn) adopting stripe ordered structures, and is supported by the observation that the Mn, Fe and Co cations are under bonded in the MCaTa₂O₇ phases (Mn BVS = +1.85; Fe BVS = +1.81; Co BVS = +1.80) while the bonding of the ZnO₄ units in ZnCaTa₂O₇ has been optimized (Zn BVS = +2.01, + 1.99). Thus, it can be seen that the size of the M ²⁺ interlayer cations determines the relative orientation of the tilting distortions of adjacent CaTa₂O₇ layers.

FeCaTa₂O₇, CoCaTa₂O₇, and isostructural MnCaTa₂O₇ adopt analogous magnetically ordered structures at low temperature, with ordering temperatures which scale with the size of the local moment (CoCaTa₂O₇: T _N = 25K, Co²⁺ S = ³/₂ ; FeCaTa₂O₇: T _N = 40K, Fe²⁺ S = 2; MnCaTa₂O₇: T _N = 56K, Mn²⁺ S = ⁵/₂). This ‘G-type’ antiferromagnetic order indicates that nearest neighbor antiferromagnetic couplings are the dominant interaction. The signatures of a glassy component to the low-temperature magnetic state of FeCaTa₂O₇ can be explained by observing that, in a tetrahedral coordination, Fe²⁺ has a nonspherical e³t₂ ³ electronic configuration (compared to spherical Co²⁺ e⁴t₂ ³ and Mn²⁺ e²t₂ ³) which could lead to the presence of some ferromagnetic nearest-neighbor couplings which will compete with, and partially frustrate, the dominant antiferromagnetic interactions.

A key reason to prepare cation-exchanged MAB ₂O₇ phases which contain paramagnetic M-cations within polar AB ₂O₇ frameworks, is to study the coupling between the magnetic and electrical polarizations in these materials which often becomes apparent at the magnetic ordering temperatures of the phases. For example, MnSrTa₂O₇ and MnCaTa₂O₇ exhibit large lattice parameter anomalies around 5 K below their respective Néel temperatures (referred to as T _A) which are associated with a step change (MnSrTa₂O₇) or local maximum (MnCaTa₂O₇) in the Γ₅ ^– polar distortion mode of the phases and are taken as indications of coupling between the magnetic and electrical polarizations present. , Figure shows plots of the lattice parameters of FeCaTa₂O₇ and CoCaTa₂O₇ as a function of temperature around their respective Néel temperatures. Neither set of data show anomalies analogous to those seen in the MnATa₂O₇ phases. Furthermore, if the magnitudes of the X₂ ⁺, X₃ ^– and Γ₅ ^– distortion modes (required to obtain the P2₁ nm symmetry structures of FeCaTa₂O₇ and CoCaTa₂O₇ from an aristotype I4/mmm phase) are plotted as a function of temperature over the same ranges (Figures S9 and S10) these also show no significant anomalies, and thus provide no evidence for magnetoelectric coupling in FeCaTa₂O₇ or CoCaTa₂O₇.

6.

Lattice parameters plotted as a function of temperature for (top) FeCaTa₂O₇ and (bottom) CoCaTa₂O₇. Dashed lines indicate the Néel temperatures of the two phases.

Conclusion

FeCaTa₂O₇, CoCaTa₂O₇ and ZnCaTa₂O₇ adopt polar crystal structures consistent with the trilinear-coupled hybrid-improper stabilization mechanism. FeCaTa₂O₇ and CoCaTa₂O₇ adopt structures described in space group P2₁ nm. Comparison of these structures to an I4/mmm, transition-metal-disordered aristotype phase reveals they are related by the application 4 symmetry lowering distortions with significant magnitude: M₂ ⁺ (chequerboard cation order); X₂ ⁺(0; a) (a ⁰ a ⁰ c ⁺/ a ⁰ a ⁰ c ⁺ tilt); X₃ ^–(b; c) (a ^– b ^– c ⁰/ b ^– a ^– c ⁰ tilt); Γ₅ ^– (polar distortion), with the presence of the X₂ ⁺(0; a), X₃ ^–(b; c) and Γ₅ ^– modes consistent with the trilinear-coupled hybrid-improper stabilization mechanism. In contrast, ZnCaTa₂O₇ adopts a crystal structure described in space group P2cm which is related to a Zn-disordered, I4/mmm symmetry aristotype structure by the application only 3 symmetry lowering distortions with significant magnitude: X₄ ^– (a, b) (combined Zn stripe-order and a ^– b ^– c ⁰/-(a ^– b ^–)c ⁰ tilt); X₂ ⁺(a; 0) (a ⁰ a ⁰ c ⁺/ a ⁰ a ⁰ c ⁺ tilt) and Γ₅ ^– (polar distortion) which are also symmetry compatible with a trilinear coupling hybrid-improper stabilization mechanism for the observed polar structure.

FeCaTa₂O₇ and CoCaTa₂O₇ adopt ordered antiferromagnetic states at low temperature with a G-type arrangement, directly analogous to that of MnCaTa₂O₇. However, in contrast to the Mn phase, neither FeCaTa₂O₇ nor CoCaTa₂O₇ show lattice parameter anomalies close to their respective Néel temperatures, and thus show no evidence for magnetoelectric coupling. This lack of magnetoelectric coupling is puzzling given the behavior of the related phases MnCaTa₂O₇ and MnSrTa₂O₇, and suggests the interactions between the electrical and magnetic polarizations in these phases are subtle and may depend on a further, as yet unidentified parameter.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.inorgchem.5c01951.

• SHG data from all samples. Complete details of refined structures of ZnCaTa₂O₇, FeCaTa₂O₇ and CoCaTa₂O₇ including selected bond lengths. AC susceptibility of FeCaTa₂O₇ and magnetization-filed isotherms of CoCaTa₂O₇ (PDF)

The manuscript was written through contributions of all authors.

The authors declare no competing financial interest.