betaFeedback

WorksPhysics and Astronomy

Universal Polar Instability In Highly Orthorhombic Perovskites

Abstract

The design of novel multiferroic ABO3_3 perovskites is complicated by the presence of necessary magnetic cations and ubiquitous antiferrodistortive modes, both of which suppress polar distortions. Using first-principles simulations, we observe that the existence of quadlinear and trilinear invariants in the free energy, coupling tilts and antipolar motions of the A and B sites to the polar mode, drives an avalanche-like transition to a non-centrosymmetric Pna21Pna2_1 symmetry in a wide range of magnetic perovskites with small tolerance factors - overcoming the above restrictions. We find that the Pna21Pna2_1 phase is especially favoured with tensile epitaxial strain, leading to an unexpected but technologically useful out-of-plane polarization. We use this mechanism to predict various novel multiferroics displaying interesting magnetoelectric properties with small polarization switching barriers.

Perovskite materials with the ABX3\mathrm{ABX}_3 chemical formula provide a fascinating playground for exploring the physics of transition metal compounds. Diverse phenomena including non-collinear magnetism, metal-insulator transitions, (anti-)ferroelectricity and superconductivity are known to exists in the structure type. Such a wide range of physical phenomena is enabled by the ability of the structure to distort - largely through tilts and rotations of the B-site octahedra - to accomodate almost any element on the A and B sites. For a given ABX3\mathrm{ABX}_3 composition, the tendency towards distortion is phenomenologically described by the tolerance factor,

t=rA+rX2(rB+rX)(1)t = \frac{r_A + r_X}{\sqrt{2}(r_B + r_X)} \tag*{(1)}

where rAr_A, rBr_B and rXr_X are the ionic radii of the A, B and X sites respectively. For small variations from the ideal t=1t = 1, octahedral distortions occur to alleviate the size mismatch. For very small or very large tt (typically t<0.7t < 0.7 or t>1.0t > 1.0), distortions can not stabilize the perovskite structure and other structural polymorphs are favoured. Such structural and chemical flexibility make perovskites ideal for engineering desired physical properties. However, engineering multiferroism (the simultaneous ordering of both electric and magnetic dipoles which can be reversed by applied external fields) with a strong coupling between the constituent electric and magnetic dipole orders has proved challenging. There are two main reasons for this: 1) polar distortions in perovskites are typically caused by the formation of bonds between B and X sites [1]. If the B-site possesses the localized dd electrons necessary for long-range magnetic ordering, such bonds are suppressed by the d0d^0 criterion [2] and 2) the octahedral tilts necessary to accommodate a wide range of cations suppress polar distortions [3, 4]. Despite the limitations enforced by the above phenomena, various mechanisms have been identified that result in the coexistence of ferromagnetism and ferroelectricity in a perovskite architecture. For example, lone-pair cations on the A-site separate the magnetic and polar modes onto distinct cation sublattices [5, 6] whilst layered perovskites can break centrosymmetry and allow ferroelectricity in a hybrid improper mechanism [7, 8]. Alternatively, the strong coupling of epitaxial strain to the polar Γ\Gamma-point phonon [9] can result in strain-stabilized non-centrosymmetric phases. This distortion is even seen to occur in systems possessing dd electrons and thus provides a mechanism to circumvent the tendency of dd systems to form centrosymmetric structures [10, 11]. In this paper, we use first principles simulations to explore an alternative mechanism to engineer multiferroicity in a wide range of perovskites despite the presence of magnetic cations and large octahedral tilts. Specifically, we observe that couplings of the polar distortion and antipolar motions of B-site cations to various antiferrodistortive modes such as octahedral rotations stabilizes an out-of-plane polarization through an unusual avalanche-related mechanism [12]. Furthermore, this mechanism appears to be universal to all perovskites with large tilting and does not rely on lone-pairs or d0d^0 ions (or layering to break centrosymmetry). While not fundamentally reliant on tensile strain, the mechanism is substantially enhanced by its application due to how strain influences the antiferrodistortive modes present in the couplings. Unexpectedly, tensile strain leads to strong out-of-plane polarizations in sharp distinction to previous studies of strain-induced ferroelectricity [13]. Finally, we show how these principles can be used to create promising multiferroics with strong spin-phonon couplings and illustrate these effects on a few candidate materials.

Results

Origins Of Pna21Pna2_1 Instability

Previous work [14, 15, 16, 17, 18, 19] has demonstrated that perovskites with a low tolerance factor can be stabilised under high-pressure or thin film synthesis techniques and very often form GdFeO3\mathrm{GdFeO}_3 type structures with a $P nma symmetry. This phase consists of two primary distortions from the cubic Pm3ˉmPm\bar{3}m reference - equal antiphase tilts about two of the axes of the octahedra and an in phase tilt of a different magnitude about the final axis, a tilt pattern described symbolically in Glazer notation as a−b+a−a^{-}b^{+}a^{-} [20]. This results in a cell that is 2×2×2\sqrt{2}\times2\times\sqrt{2} larger than Pm3ˉmPm\bar{3}m.

We explore instabilities of the PbnmaPbnma phase in a wide range of perovskite materials (see Table S1 and caption for details) as a function of epitaxial strain using first-principles simulations (see Methods). We choose to disentangle the effects of lone-pair stereochemistry and magnetically driven symmetry lowering by choosing materials in which the A-site does not possess lone pairs and is also non-magnetic (ff electrons were frozen to the core for rare earth cations). To demonstrate the generality of the effect, we also chose to study a selection of chemical compositions covering elements from the ss−, pp−, dd− and ff− blocks with a range of formal valences and dd occupancies.

MaterialttMagnetismaIPa_{\mathrm{IP}} (Å)Relaxed Pna21Pna2_1 X5+X_5^+ (Å)Relaxed Pna21Pna2_1 Γ4−\Gamma_4^- (Å)
ScCrO3_30.7539C-type AFM3.6780.5490.854
ScFeO3_30.7426G-type AFM3.6700.8141.172
ScGaO3_30.7520NM3.6590.7061.029
InCrO3_30.7737C-type AFM3.7410.7280.990
InFeO3_30.7620G-type AFM3.7340.8991.152
InGaO3_30.7717NM3.7150.8471.121
MgMnO3_30.7786FM3.5740.7151.164
MgGeO3_30.7786NM3.5730.6841.147
ZnMnO3_30.7861FM3.5770.8151.055
ZnGeO3_30.7861NM3.5680.7320.997
LuMoO3_30.7664G-type AFM3.9070.7661.087
YbMoO3_30.7688G-type AFM3.9180.6570.947

Table S1. Parameters of materials studied. Tolerance factor tt measured using 6-coordinated A-sites, giving consistency with [14]. In plane lattice constant aIPa_{\mathrm{IP}} taken as the average bulk PnmaPnma in-plane lattice constants divided by 2\sqrt{2}. Amplitudes of Pna21Pna2_1 mode, with respect to the PnmaPnma supercell, calculated at 4% strain for non rare-earth materials. For rare-earth materials, the calculation is performed at 2% strain - approximately the minimum of the PnmaPnma well. Magnetic structure determined by calculating the energies after bulk relaxation of A-, C-, G- and FM and selecting the lowest energy.

Despite the large antiferrodistortive modes present in these highly orthorhombic materials, which are known to strongly suppress any polar distortions [4], the first and third rows of Figure 1 explicitly show that a prominent instability to a polar Pna21Pna2_{1} phase is present for many perovskite oxides with low-tt. The same rows in Figure 1 also demonstrate that the relative phase stability between PbnmaPbnma, Pna21Pna2_{1} and rhombohedral LiNbO3_{3}-type R3cR3c (an extremely distorted phase typically formed when t<0.8t < 0.8 and characterised by a−a−a−a^{-}a^{-}a^{-} tilts) can be controlled via strain. We do not consider the ilmenite (Rˉ3\bar{R}3) structure due to previous reports [21] indicating that such a structure is higher in energy than the LiNbO3_{3}-type structure and this is supported in our calculations using InFeO3_{3} as a representative example - see caption to Table S2. Due to the epitaxial constraint forcing the matching of lattice vectors to the substrate, no hexagonal perovskite phases were considered. The Pmc21Pmc2_{1} and Pmn21Pmn2_{1} symmetries with polarization within the plane were not able to be stabilised under geometry relaxation [11, 22].

Energy and mode amplitudes versus epitaxial strain for oxide perovskites

Figure 1. First and third rows : energy of the Pna21Pna2_1 (orange dashed) and R3cR3c (green dash-dot) phases, relative to PnmaPnma (blue solid), vs epitaxial strain for a variety of oxide perovskites with small tolerance factors. Second and fourth rows : Total R4−R_4^- amplitude as well as the R4−R_4^- amplitude decomposed into the components affecting the A and O sites individually. Details of simulations and lattice constants can be found in Table S1.

MaterialStrain (%)Polarization (μC/cm2\mu\mathrm{C}/\mathrm{cm}^2)Magnetic StructureEasy AxisBand Gap (eV)ΔE0\Delta E_0 (meV/f.u)ΔER\Delta E_R (meV/f.u)
InCrO3_3111.15FM[100]1.8429.70317.40
InFeO3_3022.42G-AFM*[1]1.8128.30252.96
MgMnO3_3015.49FM[1]0.8513.95331.01

Table S2. Properties of candidate materials at specified strains. InFeO3_3 is a G-type antiferromagnetic - the asterisk denotes a magnetic point group allowing for a wFM moment. This has a calculated magnitude of 0.021μB\mu_B. ΔE0\Delta E_0 and ΔER\Delta E_R are the difference in energies between the two orthorhombic structures (PnmaPnma and Pna21Pna2_1) and the two rhombohedral structures (R3ˉcR\bar{3}c and R3cR3c) respectively. We use these values as a proxy for ferroelectric switching barrier height. Note that the ilmenite phase is not considered in the current study. Calculations on this phase in InFeO3_3 reveal the ilmenite structure to be 502 meV/f.u higher in energy than the PnmaPnma phase.

All materials studied in Figure 1 show a region of strain in which polar Pna21Pna2_{1} is stabilised over centrosymmetric PbnmaPbnma. For many materials, this region includes 0%, indicating that strain is not strictly necessary to stabilise Pna21Pna2_{1}. However, there exists an optimum tensile strain at which the energy difference between Pna21Pna2_{1} and PbnmaPbnma is maximised. Beyond this characteristic optimum strain, the energy difference is reduced until eventually the polarisation is destroyed. This can be seen explicitly for ScCrO3_{3}, LuMoO3_{3} and YbMoO3_{3} and is expected for the rest of the Sc and In series if higher strains were explored. However, the same behaviour is not observed in the Mg and Zn series - we explain the diversity of responses to strain later in the section.

In addition, we observe that the LiNbO3_{3}-type R3cR3c phase (or CcCc under biaxial strain) is strongly favoured by tensile strain and very rapidly becomes the stable structural polymorph. Nevertheless, the InBO3_{3} and RRMoO3_{3} families all have a window of stability for the Pna21Pna2_{1} phase at relatively low strains. Mg(Mn,Ge)O3_{3} and ZnMnO3_{3} also have a region of stability but exclusively in the compressive regime. Note that we also ran calculations on ZnSnO3_{3} (see Figure S1), observing similar trends but an R3cR3c ground state at all strains (unlike the results of Ref [23]).

Relative stability of Pnma, Pna21, and R3c phases in ZnSnO3 as a function of epitaxial mismatch strain

Figure S1. Relative stability, as a function of strain, of PnmaPnma, Pna21Pna2_1, and R3cR3c phases in ZnSnO3_3. We observe that R3cR3c is the ground state for all values of strain, in contrast to the results of [23].

Two questions are presented by the energy-strain plots in Figure 1: what mechanism is responsible for a universal polar distortion in strongly distorted perovskites and what causes the minima in the energy-strain graph? We begin our discussion by investigating the latter. To do this, we explore how the symmetry-adapted modes of the PbnmaPbnma structure are altered with tensile strain. A representative example is given for InCrO3_{3} in Figure S2. Whereas the antipolar distortion of the A sites alone (X5−X^{-}_{5}) and the average of the two tilt modes (M2+M^{+}_{2} and R5−R^{-}_{5}) remains approximately constant, we observe a pronounced minimum in the small antipolar motion of the A and O sites (R4−R^{-}_{4}). Of all the modes, tensile epitaxial strain causes the largest relative change in R4−R^{-}_{4}. In Figure S2, we show that the R4−R^{-}_{4} mode can be decomposed into two components that affect only the A and O-sites respectively. Strain causes continuous but opposite changes in both components, with each changing sign, so that there exists a minimum in the total mode amplitude. Figure S2 also shows that this mode minimum neatly overlaps with the minimum of the energy of the Pna21Pna2_{1} symmetry.

Amplitudes of modes and energy with respect to epitaxial strain

Figure S2. a) Amplitude of modes in PnmaPnma InCrO3_3. We see a pronounced minimum in the amplitude of the R4−R_4^- mode b) Minimum in R4−R_4^- is caused by opposite trends in its two components. Minimum of R4−R_4^- also coincides closely with the minimum of the Pna21Pna2_1 energy.

The lowest order coupling terms allowed by symmetry between antiferrodistortive (AFD) modes (such as octahedral tilts and antipolar motions) and polar modes (Γ4−\Gamma^{-}_{4}) have the form

ΔEeven=∑iai,Γ4−Qi2QΓ4−2(2)\Delta E_{\mathrm{even}} = \sum_i a_{i,\Gamma^{-}_{4}} Q_i^2 Q_{\Gamma^{-}_{4}}^2 \tag*{(2)}

where ii iterates over the AFD modes present in PbnmaPbnma and QΓ4−Q_{\Gamma^{-}_{4}} is the amplitude of the polar distortion. Benedek and Fennie showed that the coupling constants ai,Γ4−a_{i,\Gamma^{-}_{4}} are large and positive, leading to a suppression of the polar mode [4]. In particular, their work highlights the small R4−R^{-}_{4} as having a strong, competitive interaction with the polar mode despite its relatively small size. The materials studied in Figure 1 extend their work to perovskites with even lower tt, increasing the amplitudes of all AFD modes and subsequently also the importance of the competitive biquadratic interaction. The surprising behaviour of the R4−R^{-}_{4} with tensile strain initially leads to a reduction in this interaction, softening the polar mode just enough to lead to a stable Pna21Pna2_{1} phase. With additional increase of strain, the R4−R^{-}_{4} mode then increases leading to the hardening of the polar mode and the eventual loss of metastability for the Pna21Pna2_{1} structure.

Figure 1 shows identical behaviour in all compounds studied; the minimum of R4−R^{-}_{4} coincides with the stability of the Pna21Pna2_{1} phase, supporting the importance of the biquadratic coupling between R4−R^{-}_{4} and the polar mode. The exception is for the Zn and Mg series where the energy does not have a minima in the range of strains studied but the R4−R^{-}_{4} does. We attribute the lack of an energy minima to the slightly larger tolerance factor in these materials which leads to an overall reduction in all PnmaPnma mode magnitudes and a subsequent decrease in all biquadratic interactions.

Having explored how the lowest order even terms in the Landau expansion lead to the presence of the Pna21Pna2_1 energy minima, we note that the inclusion of higher order even terms do not add anything substantial to the analysis. Even-order terms that are quadratic in the polar mode and quartic in the tilts have been calculated to have negative coefficients so that extremely large tilts actually favour polarization in certain perovskites with R3cR3c symmetry [24]. We show explicitly in Figure S3 that this is not the case for Pna21Pna2_1.

Image not extracted

Figure S3. Plot of the amplitude of the polar distortion of 1% epitaxially strained InFeO3_3. Colour scale denotes the amplitude of the polar distortion, defined as the value of QΓ4−Q_{\Gamma_4^-} at which the energy is minimised. The deep blue colour for large tilts indicates that the polar mode is not stable when the tilts become large.

To address the second question and understand the origin of the instability, we performed further first principles simulations in which we took the fully relaxed PnmaPnma structure at a particular strain (2% for the rare earth compounds and 4% for the others) and investigated how the energy changes as we introduce structural distortions that break the symmetry to Pna21Pna2_1. These are the polar mode along the long axis (Γ4−\Gamma_4^-) and the antipolar motion of cations on the BB sites (X5+X_5^+). We have not included the other antipolar BB site mode (R5+R_5^+) introduced in the Pna21Pna2_1 symmetry because the amplitude in the fully relaxed structures is negligibly small. These results are shown in Figure 2.

Energy wells formed by scaling combinations of modes

Figure 2. Energy wells formed by scaling combinations of the modes present in Pna21Pna2_1 on top of a fully relaxed PnmaPnma structure. Modes are scaled via a parameter λ\lambda. For example, QR4=λQR40Q_{R_4} = \lambda Q_{R_4}^{0} where QR40Q_{R_4}^{0} is the relaxed amplitude of the R4−R_4^- at 4% strain (2% for rare-earth materials). A double well (minima indicated by black points) is formed when both modes are introduced.

Surprisingly, we see that neither the polar Γ4−\Gamma_4^- mode nor the antipolar X5+X_5^+ are unstable when introduced alone. If the two modes are instead introduced together, a double well forms and both modes obtain a non-zero amplitude. It is apparent that both modes must be present in order to produce a polar structure. This idea was further validated by calculating the dynamical force constants of the PnmaPnma phase in InFeO3_3 and noticing the existence of an unstable phonon of nearly 50:50 hybrid character. These results also indicate that neither polar Γ4−\Gamma_4^- or antipolar X5+X_5^+ modes are the primary order parameters. Hence, ferroelectricity can not be of proper (unstable Γ4−\Gamma_4^-) type, and indeed we find nominal Born-effective charges as was also observed in Pna21Pna2_1 fluorides [25]. Nor can ferroelectricity be of the improper (unstable X5+X_5^+ driving secondary appearance of Γ4−\Gamma_4^-) type.

We also rule out the possibility that a negative biquadratic coupling constant between X5+X_5^+ and Γ4−\Gamma_4^- is driving the simultaneous appearance of X5+X_5^+ and Γ4−\Gamma_4^-. In Figure S4, we show that this constant is actually positive. We therefore conclude that a triggered mechanism [26] caused by a cooperative biquadratic interaction is not driving the transition to Pna21Pna2_1.

Energy as a function of $Q_{\Gamma_{4}^{-}}$, showing red and blue curves for $Q_{X_{5}^{+}}=0.82$ and $Q_{X_{5}^{+}}=0.00$

Figure S4. By introducing the polar Γ4−\Gamma_{4}^{-} mode to the cubic Pm3ˉmPm\bar{3}m perovskite cell whilst simultaneously including a fixed magnitude of an antipolar X5+X_{5}^{+} distortion, we demonstrate the positive biquadratic coupling between QΓ4−Q_{\Gamma_{4}^{-}} and QX5+Q_{X_{5}^{+}}. A positive biquadratic coupling eliminates the possibility of a triggered-like ferroelectric mechanism between these two modes.

To explore other possible origins for the Pna21Pna2_1 instability, we enumerated the lowest odd order couplings in the Landau-like expansion. Terms in which both polar R4−R_4^- and the antipolar X5+X_5^+ are coupled at odd order to modes in the PnmaPnma structure would explain the unusual behaviour of Figure 2, irrespective of the sign of the coefficient. The lowest order terms of this form are

ΔEodd=c1QX5+QX5−QR4−+c2QX5+QR5−QM2+QR4−+c3QX5+QR4−QM2+QR4−+c4QX5+QR5−QM3+QR4−.\Delta E_{\mathrm{odd}} = c_1 Q_{X_5^+}Q_{X_5^-}Q_{R_4^-} + c_2 Q_{X_5^+}Q_{R_5^-}Q_{M_2^+}Q_{R_4^-} + c_3 Q_{X_5^+}Q_{R_4^-}Q_{M_2^+}Q_{R_4^-} + c_4 Q_{X_5^+}Q_{R_5^-}Q_{M_3^+}Q_{R_4^-}.

Some of these couplings were previously identified in a computational study predicting the Pna21Pna2_1 symmetry in PbCoO3_3 [27]. The terms above describe how couplings with the antipolar motion of the A-site (X5−X_5^-), the octahedral tilts (R5−R_5^- and M2+M_2^+) and another antipolar mode affecting both the A-site and O-sites (R4−R_4^-), all of which have substantial amplitude in small tolerance factor perovskites, can drive the simultaneous appearance of both R4−R_4^- and X5+X_5^+.

Behaviour of this sort is reminiscent of the avalanche-ferroelectric mechanism studied in Aurivillius compounds[12, 28, 29]. Here, the condensation of a single mode forces the condensation of two others. In the case of the present small-tt perovskites, the large amplitudes of the PnmaPnma modes in the above trilinear and quadrilinear couplings enable the condensation of both the polar mode and the antipolar X5+X_5^+ mode. Despite the mathematical similarity, whether these low tolerance factor perovskites should be classified as resulting from an avalanche transition is not clear. This classification would require that the transition goes directly from the high symmetry cubic Pm3ˉmPm\bar{3}m to the low-symmetry Pna21Pna2_1, While this might occur under certain conditions (e.g. at a particular strain and composition), it seems more likely that the octahedral tilts would condense first to produce PnmaPnma and then at a lower temperature, driven by the above couplings, the Pna21Pna2_1 phase is stabilised, though this is beyond the scope of this study. This second phase transition could be described as triggered-like since the trilinear (and quadrilinear) couplings can renormalize the quadratic terms [28, 30].

To position our work in the context of the existing literature, the Pna21Pna2_1 phase is relatively rare in perovskite oxides and there is currently much debate into the causes. The phase has been observed in lone-pair systems like BiInO3_3 [31], PbRuO3_3 [32] and predicted in PbCoO3_3 [27]. It has similarly been identified in various d0d^0 materials like CdTiO3_3 [23, 33, 34]. Pna21Pna2_1 symmetry has also been observed in rare earth orthoferrites and orthochromates [35, 36] but is usually ascribed to a spin-driven symmetry breaking, although conflicting reports of the Pna21Pna2_1 symmetry appearing at a much higher temperature than the rare earth TNT_N also exist [37]. While lone-pair, d0d^0 and spin-driven effects might be important in certain cases, we argue that the mechanism presented in the present study must also be present and is universal to all low-tt systems since it depends solely on the symmetry of the parent phase.

We conclude that PnmaPnma materials are driven to the Pna21Pna2_1 symmetry by odd order terms such as the trilinear and quadrilinear terms discussed above. However, these terms favouring Pna21Pna2_1 are in contest with the even-order biquadratic terms which inhibit any polarization. Tensile strain manages to reduce the amplitude of the R4−R_4^- mode so that its corresponding competitive interaction with the polar modes is lessened and the odd-order terms dominate producing the Pna21Pna2_1 symmetry. This promotion of an out-of-plane polarization in the Pna21Pna2_1 phase with tensile in-plane strain is opposite to the usual trend found in perovskites which would favour in-plane and disfavour out-of-plane polarization. Further strain starts to increase the magnitude of the R4−R_4^- mode leading to the destruction of the polar phase. The crucial contribution of the R4−R_4^- mode is particularly pronounced in small tolerance factor perovskites because this mode is larger in these materials, reaching R4−R_4^- amplitudes of 0.3 Å or more, considerably larger than the amplitudes of around 0.2 Å or less achieved in the higher tt Mg and Zn series or those explored in previous studies [4].

Candidate Multiferroic Systems

The Pna21Pna2_1 phase appears to be stable despite several of the materials under investigation being magnetic, allowing for the design of new multiferroics. Figure 3 shows the effect of magnetic ordering on the relative energies of the PnmaPnma, Pna21Pna2_1 and the R3cR3c phases as a function of epitaxial strain for three candidate multiferroics: InCrO3_3, InFeO3_3 and MgMnO3_3.

Strain control of magnetic and crystal structure in InCrO3, InFeO3 and MgMnO3

Figure 3. Strain control of magnetic and crystal structure in a) InCrO3_3, b) InFeO3_3 and c) MgMnO3_3. Solid line = PnmaPnma. Dashed line = Pna21Pna2_1. Dash-dot line = R3cR3c. Energies with respect to PnmaPnma G-type AFM.

For InCrO3_3, we find the ground state PnmaPnma magnetic order to be C-type, in agreement with experiment [15, 16]. Interestingly, the critical strain at which PnmaPnma transitions to Pna21Pna2_1 changes substantially depending on which magnetic structure is used. For G- and C-type magnetic structures, which have antiferromagnetic spins within each layer of the perovskite cell, the critical strain is approximately 0%. For A-type and ferromagnetically aligned spins, the critical strain is instead around −1%. We have seen in Section I A, that the antipolar motion at the B site contributes to the quadrilinear and trilinear invariants. The difference in critical strain could be due to a spin-phonon effect in which the ferromagnetic alignment of intralayer spins softens the antipolar B site motion. Due to this spin-phonon coupling, it should be possible to also engineer the Pna21Pna2_1 phase with a compressive strain under an externally applied magnetic field.

Owing to this lower critical strain, ferromagnetism is the magnetic ground state in the region where Pna21Pna2_1 is stable. Investigating Pna21Pna2_1 at 1% strain, reveals a band gap of EgE_g = 1.84 eV - a rare ferromagnetic insulator. At 1% strain, the polarization is 11.6 μ\muC/cm2^2 and the energy difference between the two ferromagnetic orthorhombic structures is ΔEO\Delta E_O = 29.7 meV/f.u whilst ΔER\Delta E_R = 317.4 meV/f.u between the two rhombohedral (R3ˉcR\bar{3}c and R3cR3c) structures. We use these values as a proxy for switching barrier height and predict that the orthorhombic structures have considerably smaller barriers than the rhombohedral materials, providing a potential method to sidestep the high barriers found in R3cR3c materials [38, 39].

Figure 3b shows the same calculation for InFeO3_3. We do not observe such prominent spin-phonon coupling (possibly due to the presence of ege_g orbitals on the Fe3+^{3+} cation), but the lower barrier present in the orthorhombic structure, presence of high TCT_C Fe3+^{3+}, weak ferromagnetic canting and the out-of-plane ferroelectricity make InFeO3_3 a potentially useful thin film multiferroic. MgMnO3_3 (Figure 3c) behaves analogously to InCrO3_3, likely due to the analogous d3d^3 filling in both, and also displays a ferromagnetic-insulating state over a substantial range of strains. A summary of all calculated material properties is included in Table S2 of the supplemental information. The supplemental information also illustrates how our results are altered by the strength of electronic correlation - the existence of a Pna21Pna2_1 instability is robust to correlation effects but the strains at which it is achieved are altered.

To conclude, through first principles calculations and group theoretical analysis, we have explored the key role played by the couplings between AFD modes in the PnmaPnma structure to polar distortions in stabilising the technologically useful Pna21Pna2_1 phase with an out-of-plane polarization. Couplings like these lead to an unusual avalanche-like transition in small tolerance factor perovskites. This mechanism appears resistant to the d0d^0 rule usually prohibiting the creation of novel multiferroic materials. Using this idea, we identify InFeO3_3 as a potential high TCT_C multiferroic with a small polarization switching barrier and sizeable wFM moment. We further identify InCrO3_3 and MgMnO3_3 as polar, ferromagnetic insulators exhibiting strong spin-phonon coupling.

Methods

All simulations are performed using density functional theory (DFT) as implemented in the Vienna Ab-Initio Software Package (VASP) Version 6.3.2 [40, 41, 42, 43]. We use the Perdew-Burke-Ernzerhof exchange correlation functional for solids (PBESol) [44]. We use a high plane wave cutoff energy of 800 eV to ensure convergence for all systems studied as well as a 7×5×77 \times5 \times7 Monkhorst-Pack kk-grid for the 2×2×2\sqrt{2} \times2 \times\sqrt{2} 20-atom supercell. Self consistent field calculations were continued until differences in energies were within a tolerance of 10−810^{-8} eV. Geometry relaxations were continued until the smallest Hellman-Feynman force was less than 10−310^{-3} eV/Å. We use projector augmented wave pseudopotentials in all our calculations. A summary of which electrons are treated as valence is presented in Table S3. To better approximate the effect of electron localization and correlation, we use the rotationally invariant formulation of the onsite Hubbard-UU parameter [45]. We use a consistent value of U=4U = 4 eV for all materials with unpaired dd electrons and U=0U = 0 eV for those without. This is also the methodology employed in previous computational screening studies [36]. The results are not qualitatively affected by the choice of UU - see supplemental information for details.

ElementValence Electrons
Sc3s23p64s23d13s^2 3p^6 4s^2 3d^1
Cr3p64s13d53p^6 4s^1 3d^5
Fe3d64p23d^6 4p^2
Ga3d104s24p13d^{10} 4s^2 4p^1
In4d105s25p14d^{10} 5s^2 5p^1
Mg3s23s^2
Mn3p63d54s23p^6 3d^5 4s^2
Ge3d104s24p23d^{10} 4s^2 4p^2
Zn3d104s23d^{10} 4s^2
Lu*5p65d16s25p^6 5d^1 6s^2
Mo4s24p64d55s14s^2 4p^6 4d^5 5s^1
Yb*5p64f16s25p^6 4f^1 6s^2
O2s22p42s^2 2p^4

Table S3. Valence electrons used in our projector augmented wave ultrasoft-pseudopotential calculations. Asterisks mark rare earth elements for which most or all of the highly localized ff electrons have been confined to the core

To apply epitaxial strain to our systems, we utilize an additional patch to VASP [46] that fixes selected lattice constants during a relaxation. This simulates epitaxial growth on a substrate for a perovskite film with many layers.

The symmetry analysis was conducted using the INVARIANTS tool of the ISOTROPY Software Suite [47, 48].

Data Availability

All data supporting the plots within this paper and other findings of this study are available from the Supplementary Information or the corresponding authors upon reasonable request.

Acknowledgements

CAMS and NCB would like to thank Helen He for helpful discussions relating to this research. This work used the Hamilton HPC service at Durham University. CAMS and NCB acknowledge the Leverhulme Trust for a research project grant (Grant No. RPG-2020-206).

Author Contributions

CAMS performed all computational work. NCB supervised the project. Both authors contributed to the writing of the manuscript.

Universal polar instability in highly orthorhombic perovskites - Supplemental Information

Cameron A.M. Scott; Nicholas C. Bristowe

Crystal and magnetic structure in InCrO3 as a function of Hubbard $U$ at 1% strain

Figure S5. Crystal and magnetic structure in InCrO3\mathrm{InCrO}_{3} as function of UU at 1% strain. Bold lines denote magnetic structure with Pnma crystal symmetry, dashed lines with Pna21_1 crystal symmetry and dashdot lines with R3c crystal symmetry. Lowest line denotes lowest energy crystal and magnetic structure for that particular choice of Hubbard-UU

Image not extracted

Figure S6. Crystal and magnetic structure of InFeO3_3 as function of UU at 0% strain. Bold lines denote magnetic structure with Pnma crystal symmetry, dashed lines with Pna21_1 crystal symmetry and dashdot lines with R3c crystal symmetry. Lowest line denotes lowest energy crystal and magnetic structure for that particular choice of Hubbard-UU

Crystal and magnetic structure of MnMnO3 as a function of U

Figure S7. Crystal and magnetic structure of MnMnO3_3 as function of UU at 0% strain.Bold lines denote magnetic structure with Pnma crystal symmetry, dashed lines with Pna21_1 crystal symmetry and dashdot lines with R3c crystal symmetry. Lowest line denotes lowest energy crystal and magnetic structure for that particular choice of Hubbard-UU

References

  1. [1]R. E. Cohen, Nature 358, 136 (1992).
  2. [2]N. A. Hill, Why are there so few magnetic ferroelectrics? (2000).
  3. [3]W. Zhong and D. Vanderbilt, Physical review letters 74, 2587 (1995).
  4. [4]N. A. Benedek and C. J. Fennie, The Journal of Physical Chemistry C 117, 13339 (2013).
  5. [5]T. Kimura, S. Kawamoto, I. Yamada, M. Azuma, M. Takano, and Y. Tokura, Physical Review B 67, 180401 (2003).
  6. [6]J. R. Teague, R. Gerson, and W. J. James, Solid State Communications 8, 1073 (1970).
  7. [7]E. Bousquet, M. Dawber, N. Stucki, C. Lichtensteiger, P. Hermet, S. Gariglio, J.-M. Triscone, and P. Ghosez, Nature 452, 732 (2008).
  8. [8]N. A. Benedek and C. J. Fennie, Physical review letters 106, 107204 (2011).
  9. [9]D. G. Schlom, L.-Q. Chen, C.-B. Eom, K. M. Rabe, S. K. Streiffer, and J.-M. Triscone, Annu. Rev. Mater. Res. 37, 589 (2007).
  10. [10]E. Bousquet and N. Spaldin, Physical Review Letters 107, 197603 (2011).
  11. [11]Y. Yang, W. Ren, M. Stengel, X. Yan, and L. Bellaiche, Physical review letters 109, 057602 (2012).
  12. [12]J. Perez-Mato, P. Blaha, K. Schwarz, M. Aroyo, D. Orobengoa, I. Etxebarria, and A. García, Physical Review B 77, 184104 (2008).
  13. [13]K. M. Rabe, Current Opinion in Solid State and Materials Science 9, 122 (2005).
  14. [14]A. A. Belik and W. Yi, Journal of Physics: Condensed Matter 26, 163201 (2014).
  15. [15]A. A. Belik, Y. Matsushita, M. Tanaka, and E. Takayama-Muromachi, Chemistry of Materials 24, 2197 (2012).
  16. [16]L. Ding, P. Manuel, D. D. Khalyavin, F. Orlandi, Y. Kumagai, F. Oba, W. Yi, and A. A. Belik, Physical Review B 95, 054432 (2017).
  17. [17]A. Bosak, A. Kamenev, I. Graboy, S. Antonov, O. Y. Gorbenko, A. Kaul, C. Doubourdieu, J. Senateur, V. Svechnikov, H. Zandbergen, et al., Thin Solid Films 400, 149 (2001).
  18. [18]P. A. Salvador, T.-D. Doan, B. Mercey, and B. Raveau, Chemistry of materials 10, 2592 (1998).
  19. [19]M.-R. Li, U. Adem, S. R. McMitchell, Z. Xu, C. I. Thomas, J. E. Warren, D. V. Giap, H. Niu, X. Wan, R. G. Palgrave, et al., Journal of the American Chemical Society 134, 3737 (2012).
  20. [20]A. M. Glazer, Acta Crystallographica Section B: Structural Crystallography and Crystal Chemistry 28, 3384 (1972).
  21. [21]W. D. Parker, J. M. Rondinelli, and S. Nakhmanson, Physical Review B 84, 245126 (2011).
  22. [22]C.-J. Eklund, C. Fennie, and K. Rabe, Physical Review B 79, 220101 (2009).
  23. [23]S. G. Kang, International Journal of Quantum Chemistry 117, e25420 (2017).
  24. [24]T. Gu, T. Scarbrough, Y. Yang, J. Íñiguez, L. Bellaiche, and H. Xiang, Physical review letters 120, 197602 (2018).
  25. [25]A. C. Garcia-Castro, A. H. Romero, and E. Bousquet, Physical Review Letters 116, 117202 (2016).
  26. [26]J. Toledano, Physical Review B 20, 1147 (1979).
  27. [27]F. Lou, W. Luo, J. Feng, H. Xiang, Physical Review B 99, 205104 (2019).
  28. [28]J. Perez-Mato, M. Aroyo, A. García, P. Blaha, K. Schwarz, J. Schweifer, and K. Parlinksi, Physical Review B 70, 214111 (2004).
  29. [29]U. Petralanda, J. Hlinka, and I. Etxebarria, Physical Review B 96, 144112 (2012).
  30. [30]P. Toledano and J.-c. Toledano, Landau Theory Of Phase Transitions, The: Application To Structural, Incommensurate, Magnetic And Liquid Crystal Systems, Vol. 3 (World Scientific Publishing Company, 1987).
  31. [31]A. A. Belik, S. Y. Stefanovich, B. I. Lazoryak, and E. Takayama-Muromachi, Chemistry of materials 18, 1964 (2006).
  32. [32]J.-G. Cheng, K. Kweon, J.-S. Zhou, J. Alonso, P.-P. Kong, Y. Liu, C. Jin, J. Wu, J.-F. Lin, S. A. Larregola, et al., Proceedings of the National Academy of Sciences 110, 20003 (2013).
  33. [33]P.-H. Sun, T. Nakamura, Y. J. Shan, Y. Inaguma, and M. Itoh, Ferroelectrics 217, 137 (1998).
  34. [34]H. Moriwaki, A. Kuwabara, C. A. Fisher, H. Taniguchi, M. Itoh, and I. Tanaka, Physical Review B 84, 104114 (2011).
  35. [35]G. Oliveira, R. Teixeira, R. Moreira, J. Correia, J. Araújo, and A. Lopes, Scientific reports 10, 4686 (2020).
  36. [36]H. J. Zhao, L. Bellaiche, and J. Íñiguez, Physical Review Materials 3, 064406 (2019).
  37. [37]S. Mishra, K. Rudrapal, R. Jana, K. P. Islam, A. Sagdeo, A. R. Chaudhuri, V. Adyam, D. Choudhury, et al., Physical Review B 107, 214104 (2023).
  38. [38]M. Ye and D. Vanderbilt, Physical Review B 93, 134303 (2016).
  39. [39]M. Ye and D. Vanderbilt, Physical Review B 95, 014105 (2017).
  40. [40]P. E. Blöchl, Physical review B 50, 17953 (1994).
  41. [41]G. Kresse and J. Furthmüller, Computational materials science 6, 15 (1996).
  42. [42]G. Kresse and J. Furthmüller, Physical review B 54, 11169 (1996).
  43. [43]G. Kresse and D. Joubert, Physical review b 59, 1758 (1999).
  44. [44]J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov, G. E. Scuseria, L. A. Constantin, X. Zhou, and K. Burke, Physical review letters 100, 136406 (2008).
  45. [45]S. L. Dudarev, G. A. Botton, S. Y. Savrasov, C. Humphreys, and A. P. Sutton, Physical Review B 57, 1505 (1998).
  46. [46]C. Xiao, VASP_OPT_AXIS, https://github.com/Chengcheng-Xiao/VASP_OPT_AXIS (2022).
  47. [47]H. T. Stokes, D. M. Hatch, and B. J. Campbell, Isodistort, iso.byu.edu. (2024).
  48. [48]B. J. Campbell, H. T. Stokes, D. E. Tanner, and D. M. Hatch, Journal of Applied Crystallography 39, 607 (2006).

Paper details

vlnc:W_uzwhcgp9
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents