Vertical Inversion Domain Walls

• Vertical inversion domain walls are interfaces separating regions with reversed order parameters, crucial in ferroelectrics, magnetics, and photonics.
• They exhibit rich internal dynamics, such as current-driven switching and Walker regime suppression, with velocities reaching up to 600 m/s.
• Their behavior underpins device applications like spintronic inverters and diodes, where defect interactions and geometry critically determine performance.

Searching arXiv for the cited work and closely related domain-wall literature to ground the article in current records. Vertical inversion domain walls are interfaces separating two states related by reversal of an order parameter, but the phrase is used with different technical meanings across condensed-matter physics, photonics, and gravitational theory. In materials science it most concretely denotes walls whose plane is oriented relative to a polar or magnetic axis, such as the neutral $180^\circ$ ferroelectric wall in wurtzite AlN that is parallel to $[0001]$ (Naudin et al., 19 Jun 2026), or the magnetic wall between $+z$ and $-z$ domains in garnet strips, historically described as a vertical inversion domain wall (Herranen et al., 2017). In spintronic nanodevices the same language is tied to current-perpendicular-to-plane actuation of head-to-head or tail-to-tail walls in magnetic tunnel junctions (Sampaio et al., 2013). Closely related usages include sliding-induced vertical ferroelectric domain boundaries in $\beta$-ZrI$_2$ (Ma et al., 2021), inversion-related photonic crystal interfaces carrying axionic hinge states (Devescovi et al., 2023), and a distinct holographic construction in which domain-wall geometries are mapped by conformal scale-factor inversion (Lima et al., 2019). Across these realizations, the domain wall is an active mesoscale object whose internal degrees of freedom, local symmetry, and coupling to defects or external drives govern mobility, conductivity, topology, and spectroscopic selection rules.

1. Terminological scope and definitions

The expression is not monolithic. In some literatures, “vertical” specifies wall orientation relative to a crystallographic polarization axis; in others, it denotes the direction of the driving current; in still others, “inversion” refers to a mapping of the domain-wall geometry itself rather than to a material interface.

Context	Technical meaning	Principal consequence
Wurtzite AlN	Neutral $180^\circ$ ferroelectric wall parallel to $[0001]$ and perpendicular to $[10\bar{1}0]$	Defects are stabilized at or near the wall and alter wall displacement barriers (Naudin et al., 19 Jun 2026)
Garnet strips	Magnetic wall separating $+z$ and $[0001]$0 domains, with 3D internal Bloch-line structure	Threshold-field instability proceeds via VBL or HBL excitations (Herranen et al., 2017)
Track-shaped MTJs	Head-to-head or tail-to-tail wall moved by current perpendicular to the film plane	Deterministic, reversible switching in $[0001]$1 ns with velocities up to $[0001]$2 m/s (Sampaio et al., 2013)
$[0001]$3-ZrI$[0001]$4	Vertical ferroelectric boundary between sliding-induced opposite polar states	Charged walls can host a quasi-two-dimensional electron gas (Ma et al., 2021)
Photonic crystals	Interface between two inversion-symmetric gapped phases with $[0001]$5 and $[0001]$6	Chiral hinge-localized light channels appear at inversion-related hinges (Devescovi et al., 2023)
Einstein-scalar gravity	Conformal inversion of the domain-wall scale factor $[0001]$7	UV/IR exchange between paired domain-wall solutions (Lima et al., 2019)

A persistent source of confusion is that “domain-wall inversion” can refer either to inversion of wall polarity during propagation, as in magnetic inverter devices, or to inversion-related symmetry between the domains themselves. These are distinct notions. The former is an operational transformation of a propagating wall; the latter is a statement about symmetry or topology of the adjoining phases.

2. Magnetic vertical inversion domain walls and their internal dynamics

In track-shaped magnetic tunnel junctions, the relevant structure is a multilayer stack PtMn / CoFe / Ru / CoFeB / MgO / NiFe / Ru, where the CoFe/Ru/CoFeB trilayer is a synthetic antiferromagnet and the NiFe layer is the free layer carrying the domain wall. The device uses a 110 nm wide arc-shaped track with straight ends that act as geometric pinning sites and create two stable wall positions. A domain wall is initialized by a saturating transverse magnetic field of about 80 mT, driven by short current pulses flowing perpendicular to the film plane, and detected resistively through tunneling magnetoresistance. The reported dynamics are unusually fast: for $[0001]$8 MA/cm$[0001]$9, $+z$0 ns and $+z$1 m/s; for $+z$2 MA/cm$+z$3, $+z$4 ns and $+z$5 m/s. More generally, the wall velocity reaches 400–600 m/s at current densities below 10 MA/cm$+z$6, enabling switching in approximately 1 ns (Sampaio et al., 2013).

The central dynamical interpretation is that efficient perpendicular spin-transfer torque is combined with a propagation distance short enough to suppress the usual Walker-averaged slowdown. In the same study, the onset of Walker dynamics in a long stripe occurs above $+z$7 MA/cm$+z$8, whereas in a limited track a geometry-dependent $+z$9 controls when the mean velocity drops from about 600 m/s to about 100 m/s. The governing picture is that the track can be shorter than the Walker forward stride, so the wall reaches the end of the track before retrograde motion develops.

Large-scale 3D micromagnetic simulations of garnet strips show the complementary regime in which the internal structure of a moving vertical inversion wall becomes the dominant degree of freedom. For a strip of width $-z$0 and thickness $-z$1 from 30 nm to 1.89 $-z$2m, the velocity–field curve has a low-field linear regime followed by a sudden drop at a threshold field due to excitation of internal wall modes. The wall width parameter is $-z$3 nm, and the steady mobility is $-z$4. For open boundary conditions and $-z$5 nm, vertical Bloch lines nucleate from a strip edge and traverse the wall; for $-z$6, they travel back and forth between edges; at intermediate thickness they become strongly deformed by surface demagnetizing fields and circulate around the wall perimeter; for sufficiently large thickness, horizontal Bloch lines dominate (Herranen et al., 2017).

In ultrathin DMI ferromagnets driven above the Walker threshold, vertical Bloch lines also mediate topological conversion. A field-driven Néel wall becomes corrugated, generates a VBL pair, undergoes local anchoring, and eventually sheds a bubble that can relax into a skyrmion. The paper formulates the sequence as above Walker field $-z$7 corrugated wall $-z$8 VBL pair generation $-z$9 domain-wall anchoring $\beta$0 bubble separation $\beta$1 skyrmion stabilization, with spin-wave radiation acting as the main dissipation channel (Jeong et al., 2024). This places VBLs at the center of a broader class of topological defect transformations, not merely as perturbations of wall mobility.

3. Domain-wall inversion as device logic: inverters and diodes

A separate spintronic usage of inversion appears in Pt/Co/AlOx trilayers, where a domain-wall inverter is a patterned element that converts one wall polarity into the opposite polarity during propagation: $\beta$2 and $\beta$3. The device geometry is an out-of-plane racetrack containing a narrow in-plane magnetic region. Because interfacial Dzyaloshinskii–Moriya interaction imposes a preferred chirality, the in-plane segment behaves as a laterally coupled chiral boundary condition rather than as a passive interruption (Luo et al., 2021).

The physical mechanism is explicitly two-step: annihilation of the incoming domain on one side of the in-plane region followed by nucleation of a reverse domain on the opposite side. Under field drive the wall collapses at one boundary of the in-plane segment and a reverse domain appears at the far side; under current drive the same logic is mediated by spin-orbit torque. The DMI is written as $\beta$4, and the damping-like spin-orbit torque field is

$\beta$5

The inversion efficiency is strongly geometry dependent. For a V-shaped in-plane region, the chiral-coupling gain competes with the wall-formation cost through a ratio estimated as $\beta$6, and experimentally the nucleation probability rises as the V angle $\beta$7 is reduced, reaching 100% at $\beta$8. By widening one side of the racetrack, an asymmetric inverter is obtained in which forward propagation is transmitted and inverted, whereas reverse propagation is pinned. The resulting device acts as a domain-wall diode (Luo et al., 2021). This usage is important because it shows that inversion can be engineered as a logic primitive even when no crystallographic inversion symmetry is involved.

4. Ferroelectric and ferroelastic realizations

In wurtzite AlN, the spontaneous polarization is along [0001], and the vertical inversion domain wall is a 180° domain wall whose plane is parallel to the polarization axis and perpendicular to [10$\beta$90]. The wall is constructed to be electrically neutral, atomically sharp, and about two atomic layers thick. Polarization is tracked by the rumpling parameter

$_2$0

which equals $_2$1 in bulk AlN. At the wall it is reduced; averaging over one half of the wall gives $_2$2, a reduction of roughly 29%. The wall formation energy is $_2$3, and the band gap is lowered by about 0.20 eV (Naudin et al., 19 Jun 2026).

First-principles calculations show that all studied defects except $_2$4 are more stable at or near the wall. The correlation between the energy preference $_2$5 and the structural descriptor $_2$6 has coefficient $_2$7. Wall motion was modeled with nudged elastic band calculations in which displacement proceeds by successive polarization reversals of [0001]-oriented atomic columns. For the pristine wall, the chain-by-chain mechanism contains 7 metastable states and 8 barriers, with total barrier $_2$8. A nitrogen vacancy raises the total barrier to $_2$9, a +4.63% increase, while an aluminum vacancy raises it to $180^\circ$0, a +51.54% increase. $180^\circ$1 is identified as the most detrimental defect for wall motion because the defect-containing atomic column cannot switch by itself (Naudin et al., 19 Jun 2026). This establishes a direct link between wall-local defect thermodynamics and ferroelectric switching kinetics.

$180^\circ$2-ZrI$180^\circ$3 realizes a different form of vertical inversion boundary. Here the out-of-plane dipole is generated by in-plane interlayer sliding of rigid I–Zr–I trilayers. Opposite polar configurations $180^\circ$4 and $180^\circ$5 are reached from the nonpolar $180^\circ$6 state by sliding along $180^\circ$7 to minima at $180^\circ$8 Å. The Berry-phase polarization is $180^\circ$9 for the constrained $[0001]$0 state and $[0001]$1 for the fully optimized $[0001]$2 phase, with switching barrier $[0001]$3 meV/u.c. Charged $[0001]$4 walls in $[0001]$5 and $[0001]$6 structures are stabilized by carrier compensation, have formation energy $[0001]$7, support a potential difference of $[0001]$8 eV, and in the head-to-head case host a quasi-two-dimensional electron gas. The estimated Debye length is $[0001]$9 nm (Ma et al., 2021).

Related ferroic wall physics appears in systems where inversion is not the order parameter of the bulk domains but is either restored or broken locally at the wall. In $[10\bar{1}0]$0, electric quadrupole SHG microscopy resolves ferro-rotational domains related by vertical mirror operations and shows that the walls are mirror-restoring and nonpolar. The walls appear as suppressed-intensity lines with a $[10\bar{1}0]$1 reduction in SHG intensity and an apparent width $[10\bar{1}0]$2 (Guo et al., 2022). In ferroelastic $[10\bar{1}0]$3, by contrast, the bulk remains centrosymmetric while the domain wall locally breaks inversion symmetry, making the bulk Raman-active $[10\bar{1}0]$4 mode weakly infrared active and enabling cross-peaks in 2D Raman–THz spectra that are attributed to both mechanical and electrical anharmonicity between $[10\bar{1}0]$5 and $[10\bar{1}0]$6 phonons (Mousavi et al., 30 Apr 2026). Together these studies show that wall-local symmetry can be lower or higher than the symmetry of the adjoining domains, and that the experimental signature depends strongly on the corresponding probe tensor.

5. Inversion-related domain walls in topological photonics

A photonic analog of the inversion-domain-wall concept is realized in a 3D gyrotropic Weyl photonic crystal gapped by supercell modulation. The modulation

$[10\bar{1}0]$7

couples Weyl nodes of opposite chirality and opens a gap, while inversion symmetry restricts the relevant phases to $[10\bar{1}0]$8 and $[10\bar{1}0]$9. These two insulating phases remain $+z$0-symmetric but are topologically distinct: $+z$1, $+z$2, and the relative invariant is $+z$3. In the layer-Chern description, the relative change corresponds to $+z$4, so the domain wall between $+z$5 and $+z$6 is an inversion-related interface across which the axion angle jumps by $+z$7 (Devescovi et al., 2023).

The interface itself is gapped by a small tilted magnetic field $+z$8 with $+z$9. Because the two sides have the same Chern vector but different $[0001]$00, the nontrivial response is not an anomalous Hall surface metal but a higher-order boundary phenomenon. Embedding a $[0001]$01 core inside a $[0001]$02 background produces a fully connected $[0001]$03-wire whose chiral channels localize at the $[0001]$04-related hinges parallel to $[0001]$05. There are four possible hinge realizations, labeled $[0001]$06, and the occupied pair can be switched by rotating the in-plane magnetic bias, while flipping $[0001]$07 reverses the group velocity (Devescovi et al., 2023). This system generalizes inversion-domain-wall physics from order-parameter reversal to relative axion topology.

6. Domain-wall internal structure, inverse behavior, and formal dualities

Domain walls also control inversion phenomena in a thermodynamic sense. In the two-dimensional dipolar frustrated Ising model,

$[0001]$08

the mean-field analysis compares the full model with extended walls, a sharp-wall model, and a two-spin-wall model. The decisive result is that the inverse symmetry breaking transition appears only when the wall has internal degrees of freedom. In the full mean-field model, and already in the minimally extended two-spin-wall model, wall broadening at higher temperature increases the entropy of the stripe phase and lowers its free energy. In the sharp-wall model the wall is merely a discontinuity, carries essentially no extra entropy, and the inverse transition is absent. The stripe width diverges as

$[0001]$09

with $[0001]$10 at $[0001]$11 and $[0001]$12 at $[0001]$13, indicating a continuous stripe-to-homogeneous transition (Velasque et al., 2014). A plausible implication is that inversion behavior in many wall-bearing systems depends less on the existence of domains per se than on the entropy and soft modes carried by the wall profile.

In a mathematically distinct usage, inversion defines a correspondence between Einstein-scalar domain-wall solutions. For static isotropic walls with metric

$[0001]$14

the conformal scale factor inversion map sends

$[0001]$15

With accompanying transformations of the scalar field and potential, this preserves the form of the Einstein equations, maps asymptotically AdS regions to the vicinity of naked singularities in special Liouville models, and exchanges UV and IR regimes of holographic renormalization-group flows. The scalar map is real only when

$[0001]$16

and the construction is a symmetry of the GPPZ flow (Lima et al., 2019). This is not a material interface, but it retains the central logic of “vertical inversion”: the domain wall is transformed by an inversion operation along its distinguished coordinate.

Taken together, these works show that vertical inversion domain walls are best understood as a family of interfaces and transformations organized around order-parameter reversal, wall orientation, or inversion-related topology rather than as a single universal object. In magnetic systems they delimit the onset of Walker-like breakdown, enable nanosecond switching, and seed skyrmion nucleation. In ferroelectrics they control switching barriers, defect segregation, charge accumulation, and local conductivity. In unconventional ferroics they can restore parent symmetry or locally violate inversion selection rules. In photonics they transmute a relative axion-angle jump into hinge-localized chiral transport. In statistical and holographic theories, they provide a compact way to expose how internal wall degrees of freedom or scale-factor inversion reorganize phase behavior and duality structure.