h-Lu0.6Sc0.4FeO3: Hexagonal Multiferroic Ferrite

• h-Lu0.6Sc0.4FeO3 is a stabilized metastable hexagonal ferrite that exhibits a polar lattice with improper ferroelectricity and coexisting weak ferromagnetism.
• Sc substitution stabilizes the hexagonal phase via effective chemical strain, leading to a trimerized lattice and decoupled ferroelectric and magnetic domains.
• Investigations reveal a noncollinear 120° antiferromagnetic ordering with observable THz magnon resonances and electromagnon features, advancing multiferroic studies.

Searching arXiv for relevant papers on h-Lu0.6Sc0.4FeO3 and closely related Lu-Sc hexagonal ferrites. arxiv_search(query="h-Lu0.6Sc0.4FeO3 Lu Sc FeO3 hexagonal ferroelectric weak ferromagnetic", max_results=10) arxiv_search: {"query":"h-Lu0.6Sc0.4FeO3 Lu Sc FeO3 hexagonal ferroelectric weak ferromagnetic", "max_results": 10} Hexagonal Lu$_{0.6}$Sc$_{0.4}$FeO$_3$ (h-Lu$_{0.6}$Sc$_{0.4}$FeO$_3$, often abbreviated h-LSFO) is a Sc-stabilized bulk form of the metastable hexagonal ferrite LuFeO$_3$. It is a single-phase multiferroic platform in which a polar hexagonal lattice, improper ferroelectricity, noncollinear antiferromagnetism, and weak ferromagnetism coexist within the same crystallographic framework. The compound has been studied as a model system for testing whether spontaneous ferroelectric polarization and the net weak-ferromagnetic moment can couple directly at the domain level, and for examining how structural topology constrains magnetoelectric behavior in hexagonal ferrites (Du et al., 2018).

1. Phase stabilization and crystallographic setting

Bulk LuFeO$_3$ does not naturally adopt the hexagonal structure under ambient synthesis conditions; in bulk it is ordinarily orthorhombic, whereas the hexagonal polymorph is metastable. Partial substitution of Sc for Lu stabilizes the hexagonal lattice by an effective chemical strain analogous to the epitaxial strain used in thin films. In the Lu$_{1-x}$Sc$_x$FeO$_{0.4}$0 series, one XRD survey reported that orthorhombic reflections decrease with increasing Sc content, hexagonal reflections emerge at $_{0.4}$1, and a pure hexagonal phase is obtained at $_{0.4}$2 and $_{0.4}$3 within XRD resolution; a later structural phase diagram placed pure hexagonal Lu$_{0.4}$4Sc$_{0.4}$5FeO$_{0.4}$6 in the interval $_{0.4}$7 (Lin et al., 2016, White et al., 2019). This places h-Lu$_{0.4}$8Sc$_{0.4}$9FeO$_3$0 at, or very near, the lower edge of the reported bulk hexagonal stability window.

The hexagonal phase is described in the same symmetry language used for hexagonal manganites. The paraelectric structure is $_3$1, and the polar ferroelectric phase is $_3$2. The transition is associated with structural trimerization, and one formulation describes it as the freezing of the three phonon modes

$_3$3

which drive the structural transition from $_3$4 to $_3$5 (Lin et al., 2016).

Single crystals of h-Lu$_3$6Sc$_3$7FeO$_3$8 have been grown by floating-zone methods and carefully annealed. The material was reported as highly cleavable, an important practical property because many of the central domain-imaging experiments were carried out on cleaved $_3$9-surfaces (Du et al., 2018).

2. Improper ferroelectricity and structural topology

The ferroelectricity of h-Lu$_{0.6}$0Sc$_{0.6}$1FeO$_{0.6}$2 is improper, or geometric, rather than proper. Polarization is not the primary lattice instability; instead it is induced by structural trimerization of the hexagonal lattice. In this framework, condensation of the $_{0.6}$3-point structural mode with $_{0.6}$4 triples the unit cell, rotates the FeO$_{0.6}$5 trigonal bipyramids, buckles the rare-earth planes, and produces polarization along the hexagonal $_{0.6}$6-axis (Du et al., 2018, Martinez et al., 29 Jul 2025).

Sc substitution is not presented as a mechanism that fundamentally changes the ferroelectric physics. Rather, it renders the bulk hexagonal phase accessible while preserving the essential geometric ferroelectric mechanism. First-principles calculations for the Sc-substituted system indicate that the multiferroic properties, including geometric ferroelectricity, remain robustly unaffected by partial Sc substitution, and the Berry-phase polarization of the half-substituted hexagonal system remains nearly unchanged from h-LuFeO$_{0.6}$7 (Lin et al., 2016).

A central consequence of the trimerized $_{0.6}$8 structure is the formation of structural antiphase boundaries and vortex-like domain topology. Around a structural vortex or antivortex core, the structural distortion angle winds by $_{0.6}$9, and the corresponding structural topological charge is

$_{0.4}$0

This structural topology is not a secondary detail. It governs the ferroelectric domain morphology and, according to the domain-coupling analysis developed for h-LSFO, also constrains which magnetic winding textures are topologically favorable (Du et al., 2018).

3. Magnetic order and multiferroic regime

The Fe$_{0.4}$1 sublattice forms a trimerized triangular network in the $_{0.4}$2 planes and supports noncollinear antiferromagnetism. For the hexagonal phase, first-principles calculations identify the noncollinear 120° Y-AFM state as the magnetic ground state. In the pure compound, the relative energies listed for h-LuFeO$_{0.4}$3 place Y-AFM below both the FM and A-AFM alternatives within the hexagonal structure, while partial Sc substitution stabilizes the hexagonal lattice without qualitatively changing that magnetic ground state (Lin et al., 2016).

At low temperature, h-Lu$_{0.4}$4Sc$_{0.4}$5FeO$_{0.4}$6 is discussed in terms of an $_{0.4}$7-type magnetic structure. In this state the noncollinear order is canted and carries a small net ferromagnetic moment $_{0.4}$8, collinear with the ferroelectric polarization $_{0.4}$9. Below about $_3$0–$_3$1 K, the Fe spins adopt this $_3$2-type structure, generating weak ferromagnetism along $_3$3 (Du et al., 2018, Martinez et al., 29 Jul 2025).

Magnetic characterization in the Lu$_3$4Sc$_3$5FeO$_3$6 series was reported most directly for $_3$7 and $_3$8. Those compositions show a high-temperature magnetic anomaly at $_3$9–$_3$0 K, a low-temperature transition at $_3$1–$_3$2 K, and a weak anomaly or spin-reorientation feature at $_3$3–$_3$4 K. The inverse susceptibility follows Curie-Weiss behavior above 450 K, with $_3$5 K for $_3$6 and $_3$7 K for $_3$8, indicating strong antiferromagnetic interactions (Lin et al., 2016). Since h-Lu$_3$9Sc$_3$0FeO$_3$1 lies within the same stabilization regime, a plausible implication is that it belongs to the same general magnetic family, though the specific series paper did not make $_3$2 a headline experimental case.

The exchange model used for the hexagonal ferrites retains an in-plane nearest-neighbor superexchange $_3$3 and an interlayer superexchange $_3$4,

$_3$5

with $_3$6 for high-spin Fe$_3$7. Within that description, Sc substitution leaves $_3$8 almost unchanged and increases $_3$9 from $_{1-x}$0 meV in h-LuFeO$_{1-x}$1 to $_{1-x}$2–$_{1-x}$3 meV in h-Lu$_{1-x}$4Sc$_{1-x}$5FeO$_{1-x}$6, consistent with stronger in-plane antiferromagnetic coupling in the stabilized hexagonal phase (Lin et al., 2016).

4. Ferroelectric and weak-ferromagnetic domain structures

The experimental signature for h-Lu$_{1-x}$7Sc$_{1-x}$8FeO$_{1-x}$9 as a domain-topological multiferroic comes from the direct comparison of ferroelectric and weak-ferromagnetic domain morphologies. Room-temperature piezoresponse force microscopy on cleaved $_x$0-surfaces revealed an intrinsic cloverleaf vortex pattern of ferroelectric domains, while low-temperature magnetic force microscopy at about 78 K revealed loop-like weak-ferromagnetic domains with a vastly larger characteristic scale (Du et al., 2018).

Order parameter	Probe and conditions	Observed domain characteristics
Ferroelectricity	PFM at room temperature on cleaved $_x$1-surfaces	Cloverleaf vortex-antivortex texture; size $_x$2–$_x$3m; nearly atomically sharp walls
Weak ferromagnetism	MFM around 78 K after thin Au coating	Large loop domains; size $_x$4m; wall thickness $_x$5–$_x$6m

The ferroelectric domain topology is closely reminiscent of hexagonal manganites. The characteristic size is approximately $_x$7–$_x$8m, and faster cooling produces smaller, more disordered domains, whereas slower cooling yields more regular micron-scale vortex patterns. The density of vortices and antivortices scales with cooling rate, but with a slope larger than expected from a simple Kibble–Zurek estimate, suggesting that extrinsic disorder and pinning are significant in the metastable hexagonal ferrite. Ferroelectricity was also corroborated by PUND-based polarization-electric field measurements showing switchable polarization loops with low leakage (Du et al., 2018).

The weak-ferromagnetic domains are morphologically distinct. They form large loops rather than cloverleaf vortices, their characteristic lateral scale is about $_x$9m, and their domain walls are unusually broad, roughly $_{0.4}$00–$_{0.4}$01m. The MFM measurements were performed after coating the surface with a thin Au layer to suppress electrostatic artifacts from ferroelectric contrast. Upon warming above the magnetic transition and cooling again, the WFM pattern in the same area changes completely, whereas in-situ PFM shows that the ferroelectric domain pattern remains essentially unchanged between room temperature and 78 K (Du et al., 2018).

This disparity in length scale, morphology, and thermal reproducibility is the experimental basis for the conclusion that ferroelectric and weak-ferromagnetic domains are decoupled in h-Lu$_{0.4}$02Sc$_{0.4}$03FeO$_{0.4}$04. The paper also reports no magnetoelectric current upon switching magnetic field, consistent with the absence of direct WFM–FE switching in this bulk system (Du et al., 2018).

5. Domain decoupling, topological charge, and coupling pathway

The central interpretive result is that the relevant coupling in h-Lu$_{0.4}$05Sc$_{0.4}$06FeO$_{0.4}$07 is not primarily between ferroelectric polarization and the out-of-plane weak-ferromagnetic moment, but between polarization and the dominant in-plane antiferromagnetic spin texture mediated by structural distortions. The domain-state variables are written compactly as

$_{0.4}$08

and the magnetoelectric coefficient along $_{0.4}$09 is denoted $_{0.4}$10 or $_{0.4}$11 (Du et al., 2018).

Two theoretical possibilities are discussed. In one case, FE domains would be directly coupled to WFM domains. In the other, FE domains would be coupled to the magnetoelectric state or in-plane spin texture, while WFM remains decoupled. The observed domain morphologies support the second case. Around a ferroelectric vortex core, the structural topological charge is

$_{0.4}$12

The magnetic order can also be assigned a winding number $_{0.4}$13. In the experimentally favored scenario, the in-plane spins rotate by $_{0.4}$14 between neighboring domains, giving

$_{0.4}$15

which matches the structural topological charge. In the alternative, theoretically possible but unobserved coupled scenario, the magnetic winding would be

$_{0.4}$16

corresponding to $_{0.4}$17 spin rotations across neighboring ferroelectric domain walls (Du et al., 2018).

The statement that the magnetic topological charge tends to be identical to the structural topological charge is important because it relocates the primary coupling pathway from $_{0.4}$18–$_{0.4}$19 locking to structure–AFM compatibility. In this picture, the trimerized lattice texture sets the topological framework, the in-plane antiferromagnetic order follows that framework, and the weak $_{0.4}$20-axis moment is a secondary canting component that does not track the ferroelectric vortex pattern. A common misconception in multiferroics is that coexistence of spontaneous polarization and spontaneous magnetization automatically implies interlocked domain switching; h-LSFO provides a counterexample in which bulk multiferroicity coexists with clear FE–WFM domain decoupling (Du et al., 2018).

6. Optical, THz, Raman, and electronic structure

Single-crystal h-Lu$_{0.4}$21Sc$_{0.4}$22FeO$_{0.4}$23 has also been investigated by THz spectroscopy, Raman scattering, ellipsometry, and DFT+eDMFT. In these measurements the material is treated as a polar $_{0.4}$24 multiferroic with noncollinear Fe-spin order, and the low-temperature $_{0.4}$25 phase is emphasized because it carries both $_{0.4}$26 and $_{0.4}$27 and breaks combined $_{0.4}$28 symmetry. In that language the phase is described as a strong altermagnetic state (Martinez et al., 29 Jul 2025).

The THz spectra show antiferromagnetic resonances at about $_{0.4}$29 THz and $_{0.4}$30 THz, as well as a lower-energy electromagnon near $_{0.4}$31 THz. The magnon modes harden as temperature increases and disappear above $_{0.4}$32 K. A particularly notable observation is that the $_{0.4}$33 THz resonance is a doublet already split at zero field by

$_{0.4}$34

with the two components active in opposite circular polarizations. This zero-field magnon dichroism was observed with both conventional circular polarization and THz vector-vortex beams. For $_{0.4}$35, the splitting grows further, and from the field dependence the Fe$_{0.4}$36 $_{0.4}$37-factor was extracted as

$_{0.4}$38

A reorientation field $_{0.4}$39 T switches the circular-polarization selection rules, interpreted as reversal of the weak-ferromagnetic canting and the remanent magnetization. At 4 K, the reorientation field is nonreciprocal with respect to propagation direction, with $_{0.4}$40 T for one direction and $_{0.4}$41 T for the opposite, and reversing the sample by 180° reverses this asymmetry (Martinez et al., 29 Jul 2025).

Raman spectroscopy provides an independent optical confirmation of the magnetic excitation spectrum. A broad magnetic peak appears at the same energy scale as the THz $_{0.4}$42 magnon, around $_{0.4}$43, in crossed-circular geometry. The phonon spectrum is otherwise comparatively conventional. Far-IR ellipsometry resolves 17 IR-active modes in total, with 10 $_{0.4}$44-axis modes and 7 in-plane modes, while Raman detects 22 modes at 5 K, including $_{0.4}$45, $_{0.4}$46, and $_{0.4}$47. Most phonons show only weak, Grüneisen-like temperature shifts from 5 to 300 K, implying relatively weak spin-phonon coupling. One exception is the low-frequency $_{0.4}$48 mode near $_{0.4}$49, which exhibits a Fano-type asymmetry fitted by

$_{0.4}$50

with $_{0.4}$51. This was interpreted as coupling between a discrete phonon and a continuum of polar excitations, likely two-phonon acoustic states rather than free carriers (Martinez et al., 29 Jul 2025).

The high-energy dielectric response was modeled using DFT+eDMFT by approximating the disordered alloy as a weighted average of LuFeO$_{0.4}$52 and ScFeO$_{0.4}$53 calculated in the experimentally determined unit cell. The strongest experimental electronic transitions occur near 4 eV for the $_{0.4}$54-axis response and 3.7 eV in-plane. The analysis assigns the band-edge transitions primarily to Fe $_{0.4}$55-O $_{0.4}$56 hybridized states, while above $_{0.4}$57 eV the response is dominated by O $_{0.4}$58-Sc $_{0.4}$59 and O $_{0.4}$60-Lu $_{0.4}$61 states. The self-energies show several Mott gaps opened by strong correlations together with one more weakly correlated band-like gap, indicating mixed Mott/covalent character. Within that treatment, the average Fe oxidation state is closer to Fe$_{0.4}$62 than to the formal Fe$_{0.4}$63, and the local moment is reduced below the high-spin $_{0.4}$64 value (Martinez et al., 29 Jul 2025).

h-Lu$_{0.4}$65Sc$_{0.4}$66FeO$_{0.4}$67 is therefore notable not simply because it combines polarization and magnetism, but because it exposes how structural trimerization, antiferromagnetic spin topology, weak ferromagnetic canting, and optical activity can coexist without requiring ferroelectric and weak-ferromagnetic domains to be spatially interlocked. The compound occupies a technically important position within the Lu$_{0.4}$68Sc$_{0.4}$69FeO$_{0.4}$70 family: Sc substitution stabilizes the bulk hexagonal phase, the improper ferroelectric mechanism remains intact, the hexagonal antiferromagnetic order persists, and the resulting single crystals support both topological-domain studies and momentum-sensitive optical spectroscopy (Du et al., 2018, Lin et al., 2016, White et al., 2019, Martinez et al., 29 Jul 2025).