Localized Magnons: Mechanisms & Experiments

• Localized magnons are spin excitations whose energy and wavefunction are confined to finite regions, contrasting with extended Bloch waves.
• They arise through multiple mechanisms including anisotropy at domain walls, interface effects, and disorder-induced interference, each yielding unique quantized spectra.
• Experimental realizations in materials like CaFe₂O₄ and twisted CrI₃ demonstrate how controlled anisotropy and topology govern magnon localization and dynamic properties.

Searching arXiv for papers on localized magnons and related mechanisms to support the article. {"query":"localized magnons antiferromagnet anisotropic domain wall confined magnons moire magnets weak localization magnons arXiv", "max_results": 10} Reviewing the most relevant arXiv results for localized magnons across experimental, topological, interfacial, and disorder-induced settings. Localized magnons are spin excitations whose energy and wavefunction are confined to a finite region of a magnetic system rather than extending as Bloch-like spin waves across the crystal. The term encompasses several distinct localization mechanisms. In anisotropic bulk magnets, localized modes can arise near domain walls or antiphase boundaries and form a discrete hierarchy of quantized levels, as demonstrated in CaFe$_2$O$_4$ (Stock et al., 2016). At surfaces and interfaces, extra boundary anisotropy can bind exchange-dominated confined magnon resonances (Beairsto et al., 2021), while in twisted bilayer magnets, spatially modulated interlayer exchange can localize magnons on domain walls or at corners through topological mechanisms (Kim et al., 2022). In disordered magnets, by contrast, “localized magnons” may refer not to strictly bound eigenmodes but to weak- or Anderson-localized spin waves arising from interference under multiple scattering (Arakawa et al., 2017, Evers et al., 2015). The subject therefore spans nonlinear classical spin dynamics, boundary-value spectral theory, topological band theory, and wave localization in random media.

1. Definition and scope

Localized magnons are distinguished from ordinary magnons by spatial confinement. Extended magnons propagate through a translationally invariant magnetic medium with a wavevector-resolved dispersion, whereas localized magnons are concentrated near a defect, surface, interface, domain wall, or other inhomogeneity. In CaFe$_2$O$_4$, the localized excitations reside near antiphase-boundary regions and appear as discrete quantized spin-wave levels rather than as a continuous dispersive branch (Stock et al., 2016). In semi-infinite ferromagnets and antiferromagnets, confinement occurs near a surface or interface and produces exponentially decaying eigenmodes along the surface-normal direction while remaining dispersive parallel to the boundary (Beairsto et al., 2021).

The term also appears in distinct but related contexts. In twisted bilayer CrI$_3$, localized magnons emerge either as domain-wall-confined one-dimensional topological edge modes in the magnetic-domain phase or as corner-localized magnons in a higher-order topological magnonic insulator realized in the large-angle collinear ferromagnetic phase (Kim et al., 2022). In disordered two-dimensional antiferromagnets and chiral magnets, localization is used in the weak-localization sense: the excitations remain extended over short scales, but coherent backscattering and Cooperon-like singularities suppress transport and enhance return probability (Arakawa et al., 2017, Evers et al., 2017). In one-dimensional disordered magnets, disorder can instead drive Anderson localization with exponentially decaying transmission and a finite localization length (Evers et al., 2015).

A common misconception is that all localized magnons arise from geometric frustration and flat-band physics. The literature summarized here shows otherwise. CaFe$_2$O$_4$ is explicitly described as unfrustrated, with localization originating from anisotropy-assisted domain-wall physics and dimensional anisotropy rather than flat-band destructive interference (Stock et al., 2016). Likewise, confined surface magnons are produced by interface anisotropy (Beairsto et al., 2021), and moiré-localized magnons are generated by spatially varying interlayer exchange and symmetry-selected mass terms (Kim et al., 2022).

2. Anisotropy, competing order, and solitary magnons in CaFe$_2$O$_4$

CaFe$_2$O$_4$0 provides an experimental realization of localized magnons in a classical $_4$1 anisotropic antiferromagnet (Stock et al., 2016). The material is orthorhombic, space group Pnma, with lattice parameters $_4$2 Å, $_4$3 Å, and $_4$4 Å. The Fe$_4$5 ions form zig-zag chains and ferromagnetic stripes oriented along $_4$6, and the ordered moments align along $_4$7 because of easy-axis anisotropy. Two unfrustrated antiferromagnetic phases compete: the high-temperature B phase with $_4$8 stacking along $_4$9, and the low-temperature A phase with $_2$0 stacking along $_2$1 (Stock et al., 2016).

The near-degeneracy of these stackings produces antiphase boundaries along $_2$2. Temperature-dependent diffraction shows that the B phase onsets near $_2$3 K, persists to low temperature, and is gradually overtaken by the A phase on cooling. At the same temperature scale, an anisotropy gap opens at the magnetic ordering temperature $_2$4 K and grows on cooling; high-resolution spectroscopy gives a low-temperature gap $_2$5 meV (Stock et al., 2016). The in-plane spin-wave velocity is $_2$6 meV Å, whereas the interplane coupling is much weaker, with $_2$7. This strong exchange anisotropy renders the magnet effectively two-dimensional in its dynamics and favors stacking faults along $_2$8.

The diffuse magnetic scattering is described by an antiphase-boundary model with long-range order along $_2$9 and short correlation length along $_4$0. The elastic cross section takes the form

$_4$1

with $_4$2 b. Fits give a $_4$3-axis correlation length $_4$4 Å, corresponding to localization over roughly $_4$5–$_4$6 $_4$7-axis lattice constants, and a diffuse ordered moment $_4$8 (Stock et al., 2016).

The most striking result is the appearance of nine discrete localized spin-wave levels below $_4$9, resolved by neutron backscattering with elastic resolution $_3$0 meV (Stock et al., 2016). At $_3$1 K the spectrum is broad and characteristic of critical scattering. At $_3$2 K, precisely when the antiphase boundaries become static, the spectrum changes to a hierarchy of sharp quantized peaks, and this discrete structure persists on further cooling. The observed level positions do not follow the Airy-function quantization expected for linear-potential spinon confinement, ruling out that mechanism. Instead, the increasing spacing with excitation index is consistent with the Landé interval rule and with semiclassical predictions for localized nonlinear modes in anisotropic classical chains (Stock et al., 2016).

The spin dynamics are modeled by

$_3$3

The data suggest that the coexistence of A and B order generates antiphase boundaries that act as magnon traps, while easy-axis anisotropy provides the nonlinearity required for solitary-magnon or discrete-breather behavior. A plausible implication is that CaFe$_3$4O$_3$5 supplies a rare case in which localized magnons are observed in a classical, high-spin, unfrustrated magnet without nanostructuring (Stock et al., 2016).

3. Boundary-confined magnons at surfaces and interfaces

A separate and analytically tractable class of localized magnons occurs at surfaces and interfaces, where translational symmetry is explicitly broken and the local anisotropy can differ from its bulk value (Beairsto et al., 2021). The relevant spin Hamiltonian for a semi-infinite crystal with interface normal $_3$6 is

$_3$7

with $_3$8 at the interface and $_3$9 in the bulk. The central control parameter is the excess boundary anisotropy $_2$0 (Beairsto et al., 2021).

The theory shows that extra surface or interface anisotropy acts as an effective boundary potential. For both ferromagnets and antiferromagnets in $_2$1, the confined modes decay exponentially away from the boundary, with confinement length $_2$2. In the ferromagnetic case,

$_2$3

and the same functional form holds for antiferromagnets after $_2$4 (Beairsto et al., 2021). Increasing positive $_2$5 strengthens confinement by increasing $_2$6, while moving away from $_2$7 can also enhance localization through the structure factor $_2$8.

The confined-mode energies differ between ferromagnets and antiferromagnets, but in both cases the branch can lie below the bulk continuum as an acoustic confined mode for $_2$9, or above it as an optical confined mode when $_4$0 is sufficiently large and positive (Beairsto et al., 2021). Stability requires that the lowest confined mode remain positive at $_4$1; otherwise interface spins undergo a spin-flip instability. The threshold is

$_4$2

A notable exception occurs in the one-dimensional antiferromagnetic chain. There, an edge-localized magnon exists even for $_4$3, with finite bound-state energy and explicit confinement length (Beairsto et al., 2021). This is qualitatively different from the corresponding ferromagnetic chain and is one of the sharpest FM–AFM contrasts in the theory.

The experimentally relevant observable is the spin dynamical structure factor,

$_4$4

with the neutron-scattering cross section projected onto transverse fluctuations by

$_4$5

Confined magnons appear as additional branches or resonances in $_4$6, dispersing with $_4$7 but decaying along the surface normal (Beairsto et al., 2021). The theory therefore supplies a general diagnostic framework for surface-sensitive probes such as SPEELS, RIXS, Raman, and THz spectroscopy.

4. Moiré and topological localization in twisted bilayer CrI$_4$8

In moiré magnets, localization is not primarily caused by isolated defects or boundary anisotropy but by spatially modulated interlayer exchange across a superlattice (Kim et al., 2022). Twisted bilayer CrI$_4$9 contains distinct local stacking regions within each moiré cell, including AB and AB$_2$0 patches with different magnetic tendencies. The interlayer exchange $_2$1 depends sensitively on local stacking and is sublattice selective because the iodine sublattice breaks $_2$2 symmetry. As the twist angle $_2$3 decreases, the system passes through ferromagnetic, non-collinear-domain, and magnetic-domain phases (Kim et al., 2022).

The spin Hamiltonian is

$_2$4

with $_2$5 the intralayer ferromagnetic exchange, $_2$6 the stacking-dependent interlayer exchange, and $_2$7 the easy-axis anisotropy (Kim et al., 2022). The continuum free energy uses a coarse-grained $_2$8,

$_2$9

Two localization mechanisms are identified. In the magnetic-domain phase, AFM patches nucleate domains, and the resulting domain walls confine one-dimensional edge magnons. When next-nearest-neighbor DMI is included, it provides a Haldane mass with opposite sign in regions of opposite spin polarization. The low-energy domain-wall theory is a Dirac Hamiltonian with a sign-changing mass,

$_4$0

which supports a Jackiw–Rebbi bound state with

$_4$1

For CrI$_4$2, the Chern-number difference across the wall is $_4$3, implying two co-propagating chiral edge magnons localized on each domain wall (Kim et al., 2022).

At large twist angles, by contrast, intervalley mixing gaps Dirac magnons in the collinear ferromagnetic phase and yields a higher-order topological magnonic insulator. The relevant topological index is expressed through mirror-resolved Zak phases,

$_4$4

The resulting zero-dimensional corner modes are localized at symmetry-related sample corners and do not require either non-collinear order or intrinsic DMI (Kim et al., 2022).

The phase structure is strongly twist-angle dependent: FM for $_4$5, NCD for $_4$6, and MD for $_4$7 (Kim et al., 2022). This makes moiré magnets unusual among localized-magnon systems in that a geometric control parameter continuously tunes domain size, wall connectivity, and topological mass structure. A plausible implication is that moiré engineering provides an in situ route from localized loop modes on closed domain walls to network-like propagating channels on connected wall structures.

5. Disorder-induced localization and coherent backscattering

Disorder produces yet another class of localized-magnon phenomena, conceptually distinct from defect-bound or topological modes. In disordered two-dimensional antiferromagnets, maximally crossed impurity-scattering processes generate a Cooperon-like singularity that enhances exact backscattering and logarithmically suppresses the longitudinal magnon thermal conductivity $_4$8 (Arakawa et al., 2017). The model Hamiltonian is

$_4$9

supplemented by impurity terms acting as sublattice-dependent Zeeman fields (Arakawa et al., 2017).

The Born contribution to thermal conductivity takes the semiclassical form

$_2$0

and the weak-localization correction in two dimensions becomes

$_2$1

with

$_2$2

Thus the hallmark is a logarithmic size dependence rather than discrete bound-state peaks (Arakawa et al., 2017).

A related picture emerges from classical-wave simulations of disordered magnets (Evers et al., 2015). In one-dimensional channels, all spin-wave states are localized for arbitrarily weak uncorrelated disorder, with transmission $_2$3 and ensemble-averaged intensity profiles decaying exponentially away from the source (Evers et al., 2015). In two-dimensional films, the numerical signature of weak localization is coherent backscattering, seen as a CBS peak at $_2$4 on the elastic shell. The contrast obeys

$_2$5

with $_2$6 in the diffusive regime (Evers et al., 2015).

In chiral magnets with Dzyaloshinskii–Moriya interaction, inversion symmetry of the magnon dispersion is shifted rather than destroyed. The clean dispersion satisfies

$_2$7

with $_2$8, and the coherent-backscattering peak is correspondingly shifted to

$_2$9

instead of $_4$00 (Evers et al., 2017). The Cooperon pole moves to $_4$01, but the CBS contrast is not reduced. This provides a direct route to extract the sign and magnitude of DMI from the momentum-space shift of the backscattering peak (Evers et al., 2017).

These disorder-based phenomena are often grouped with localized magnons, but they should be distinguished from bound states. The magnons are not necessarily trapped near a specific defect or wall; rather, transport is suppressed by wave interference. That distinction is essential when comparing disorder-induced localization to solitary magnons in CaFe$_4$02O$_4$03 or edge/corner modes in moiré magnets.

6. Experimental signatures, theoretical diagnostics, and related directions

The experimental phenomenology of localized magnons depends strongly on the localization mechanism. In CaFe$_4$04O$_4$05, the defining signatures are rod-like scattering along $_4$06, discrete low-energy peaks in $_4$07, a localization length $_4$08 Å, and dynamic freezing of antiphase boundaries as measured by neutron spin echo (Stock et al., 2016). The normalized intermediate scattering function follows a stretched-exponential form,

$_4$09

with $_4$10 ps and $_4$11, while the increasing static fraction below $_4$12 K indicates freezing on $_4$13 ns timescales (Stock et al., 2016).

For surface-confined magnons, the critical observable is the spin scattering function or related surface-sensitive spectral weight. Confined branches disperse with in-plane momentum but remain outside the bulk continuum and have weight concentrated near boundaries (Beairsto et al., 2021). In thin ferromagnetic films, boundary-induced localization can hybridize with nonreciprocity. In Ni$_4$14Fe$_4$15 films, magnetostatic surface spin waves and the first perpendicular standing spin wave hybridize into two opposite-surface-localized branches that propagate in opposite directions, with localization and nonreciprocity locked together and field-tunable (Song et al., 2020). This is not a generic localized-magnon mechanism, but it illustrates how surface confinement can combine with dipole–exchange hybridization in continuous films (Song et al., 2020).

In moiré magnets, the predicted observables include gapped or gapless Dirac-magnon spectra depending on twist angle, domain-wall midgap bands in the magnetic-domain phase, and corner-localized modes in the higher-order topological regime (Kim et al., 2022). Brillouin light scattering, inelastic neutron scattering, and NV-center magnetometry are all proposed as probes of the localized edge and corner modes (Kim et al., 2022).

Several neighboring research directions broaden the concept further. In a binary Bose–Einstein condensate, p-wave pairing between distinct bosons can produce self-localized magnons with an effective mass

$_4$16

which diverges at a critical density and can generate a magnetoroton minimum (Andreev et al., 2018). In frustrated centrosymmetric magnets, hybridization of finite-$_4$17 magnons with atomic-scale skyrmions can create real-space magnon superlattices and topological magnon bands (Hullahalli et al., 1 Jan 2026). These settings are physically remote from crystalline antiferromagnets such as CaFe$_4$18O$_4$19, but they reinforce the broader principle that magnon localization can result from nonlinearity, topology, emergent textures, boundaries, or many-body entrainment rather than from a single universal mechanism (Andreev et al., 2018, Hullahalli et al., 1 Jan 2026).

Taken together, the literature defines localized magnons as a family of confined spin excitations whose realization depends on the interplay of anisotropy, dimensionality, defects, symmetry, and coherence. In one class, exemplified by CaFe$_4$20O$_4$21, easy-axis anisotropy and competing order generate antiphase-boundary traps that host quantized solitary-magnon-like states (Stock et al., 2016). In another, interface anisotropy binds exchange-dominated surface modes (Beairsto et al., 2021). In moiré systems, spatially varying interlayer exchange and topological mass inversion generate edge and corner localization (Kim et al., 2022). In disordered magnets, coherent multiple scattering suppresses transport and yields weak- or Anderson-localized spin-wave behavior (Arakawa et al., 2017, Evers et al., 2015). The unifying concept is confinement of spin-wave amplitude; the specific physics, however, is highly mechanism dependent and should not be conflated across these distinct regimes.