Domain Wall Depinning Analysis

• The paper demonstrates that domain wall depinning analysis quantitatively assesses how magnetic walls escape from pinning sites under external drives, establishing thresholds like depinning fields and current densities.
• It utilizes diverse modeling frameworks such as micromagnetic simulations, collective-coordinate models, and scalar-field approaches to capture internal wall dynamics and shape transitions.
• Controlled experiments and device models reveal that geometric notches, disorder, damping, and other parameters critically modulate depinning thresholds and drive efficiencies.

Domain wall depinning analysis is the quantitative study of how a magnetic domain wall leaves a metastable pinning site under an external drive, and of how the threshold drive depends on geometry, disorder, internal wall structure, damping, and temperature. In the literature considered here, the drive is a magnetic field, a charge current, a pure spin current, a microwave field, or an electric field; the pinning landscape is created by quenched disorder, lithographic notches, segmented corners, Pt-induced anisotropy modification, nearby nanoparticles, or interlayer coupling; and the analysis combines micromagnetics, collective-coordinate models, elastic-interface scaling, scalar-field descriptions, and atomistic simulation (Pardo et al., 2016, Kurniawan et al., 2015, Moretti et al., 2017).

1. Pinning landscapes and what “depinning” measures

Depinning denotes the transition from a pinned or metastable wall to a state in which the wall escapes a local barrier and propagates. The threshold quantity is reported as a depinning field $H_{\mathrm{dep}}$, a depinning current density $J_d$, or, in probabilistic finite-temperature settings, as a depinning probability or depinning-time distribution. In rotating domain-wall sensors, this threshold sets the lower bound of the operating field window, below the nucleation field, so the distinction between propagation-limited switching and nucleation-limited switching is operationally central rather than semantic (Heinze et al., 2016).

The pinning landscape may be geometric, magnetostatic, anisotropy-defined, or disorder-defined. Geometric examples include symmetric notches in Permalloy nanowires, segmented corners in square-loop sensors, and periodic arrays of triangular holes, where depinning is governed by the existence or loss of admissible wall shapes satisfying the local geometric constraints (Kurniawan et al., 2015, Heinze et al., 2016, Marconi et al., 2010). Magnetostatic examples include nanowire–nanoparticle gates, where two nanoparticles create either a barrier before the gate or a well between the particles depending on their magnetic configuration (Mironov et al., 2011). Anisotropy-defined pinning appears in ferrimagnetic TmIG/Pt systems, where a Pt strip locally reduces the perpendicular anisotropy barrier and creates strong pinning at its edges (Wang et al., 21 Apr 2026). Disorder-defined pinning appears in ultrathin ferromagnets, scalar-field models, and random-field Ising descriptions, where depinning separates pinned or creep-like motion from faster regimes in a disordered energy landscape (Pardo et al., 2016, Caballero et al., 2018, Xi et al., 2015).

A recurring result is that the threshold is not determined solely by a static barrier height. In notched Permalloy wires, successful escape is tied to a transformation of the wall from a transverse wall to an anti-vortex wall (Kurniawan et al., 2015). In chiral PMA systems with DMI, the dynamic depinning field can be substantially lower than the static one because transient wall motion overshoots a finite barrier (Moretti et al., 2017). In notched antiferromagnetic nanoribbons, weak damping lowers the required staggered field because the wall coordinate itself oscillates and can overshoot the notch potential (Chen et al., 2019). This supports a broader interpretation in which depinning is a dynamical transition in a structured phase space, not only a static force-balance condition.

2. Modeling frameworks and observables

Micromagnetic simulation remains a standard framework for engineered nanostructures. OOMMF with the IBM spin-transfer-torque extension was used for current-driven depinning in notched Permalloy wires, with an adiabatic spin-transfer-torque model following the Landau–Lifshitz–Gilbert equation and no $$\beta\, \mathbf{m}\times[(\mathbf{u}\cdot\nabla)\mathbf{m}]$$ term in the equation as presented (Kurniawan et al., 2015). MuMax$^3$ was used for pinned-wall resonance and microwave-assisted depinning in TmIG/Pt structures, and for chirality-dependent depinning in CoFeB-based trilayers (Wang et al., 21 Apr 2026, Jong et al., 2020). Scalar-field and Ginzburg–Landau approaches replace the wall by the $\phi=0$ contour of a full order-parameter field and thereby retain overhangs, pinch-off loops, and non-single-valued wall geometries (Caballero et al., 2018, Kolton et al., 2023).

Collective-coordinate models reduce the wall to a small set of soft variables. In weak-drive thermally activated motion, the variables are commonly wall position $q$ and internal angle $\psi$ or $\phi$, with magnetic field and nonadiabatic spin-transfer torque entering in the same combination $H-\beta\chi J$, while adiabatic torque couples to the internal wall angle and generates corrections to depinning and creep analysis (Ryu et al., 2011). In electric-field-driven MTJ structures, the reduced wall variables are wall position $X$ and internal angle $J_d$0, with VCMA shifting the equilibrium from Néel-like to Bloch-like configurations and thereby inducing precessional translation (Upadhyaya et al., 2013). In antiferromagnetic notch depinning, the reduced coordinate obeys a damped driven oscillator equation for $J_d$1, reflecting the inertial structure of AFM wall dynamics (Chen et al., 2019).

The observables are correspondingly diverse. Velocity $J_d$2 is the main observable in ultrathin-film depinning and creep studies (Pardo et al., 2016, Caballero et al., 2017). In nanowires, wall position can be inferred from average longitudinal magnetization, or by direct imaging and resistance changes (Kurniawan et al., 2015, Ilgaz et al., 2010). Sensor studies use longitudinal MOKE microscopy to record the field at which an observed branch reverses under angular scans (Heinze et al., 2016). TmIG studies combine scanning NV magnetometry, which directly images the pinned wall at the Pt edge and gives a wall width of about $J_d$3 nm, with nonlocal spin pumping, where the disappearance of the wall resonance marks depinning (Wang et al., 21 Apr 2026). Artificial-synapse studies use string-method energy profiles, depinning probability versus current and pulse width, and the distribution of depinning times over 144 thermal realizations (Kaur et al., 25 Jan 2025).

Several definitions are especially standard. In notched Permalloy wires, $J_d$4 is the minimum current density that allows the wall to escape from the notch, $J_d$5 is the depinning time, and $J_d$6 is the onset time of anti-vortex-core nucleation (Kurniawan et al., 2015). In synapse-oriented SOT tracks, $J_d$7 at $J_d$8 is the minimum current density that destabilizes the wall from the notch center, while at $J_d$9 K the problem becomes probabilistic and is characterized by depinning probability and $$\beta\, \mathbf{m}\times[(\mathbf{u}\cdot\nabla)\mathbf{m}]$$0 statistics (Kaur et al., 25 Jan 2025).

3. Mechanisms of depinning

In current-driven notched Permalloy nanowires, depinning is strongly coupled to internal wall conversion. A head-to-head transverse wall initially pinned at a symmetric notch transforms into an anti-vortex wall during escape, with a qualitative crossover near $$\beta\, \mathbf{m}\times[(\mathbf{u}\cdot\nabla)\mathbf{m}]$$1: for smaller notches the anti-vortex wall forms and depins while the current pulse is active, whereas for larger notches the anti-vortex survives until the pulse ends and depinning occurs only after a flipped transverse wall is formed (Kurniawan et al., 2015). The reported threshold decreases as both notch size and wire width increase, although the width dependence is not perfectly monotonic at the very smallest notches.

In segmented-corner domain-wall sensors, the depinning field depends on two geometric angles rather than a single tangential-field projection. The paper models this with

$$\beta\, \mathbf{m}\times[(\mathbf{u}\cdot\nabla)\mathbf{m}]$$2

and takes the measured threshold as the maximum of the two (Heinze et al., 2016). This produces divergences at $$\beta\, \mathbf{m}\times[(\mathbf{u}\cdot\nabla)\mathbf{m}]$$3, an additional singularity near $$\beta\, \mathbf{m}\times[(\mathbf{u}\cdot\nabla)\mathbf{m}]$$4, a global minimum near $$\beta\, \mathbf{m}\times[(\mathbf{u}\cdot\nabla)\mathbf{m}]$$5, and a second local minimum near $$\beta\, \mathbf{m}\times[(\mathbf{u}\cdot\nabla)\mathbf{m}]$$6. At $$\beta\, \mathbf{m}\times[(\mathbf{u}\cdot\nabla)\mathbf{m}]$$7, switching is reported to be nucleation-limited with a measured value of about $$\beta\, \mathbf{m}\times[(\mathbf{u}\cdot\nabla)\mathbf{m}]$$8, not propagation-limited.

In ferrimagnetic TmIG, depinning can be driven coherently through a localized domain-wall mode inside the magnon gap. Under weak microwave drive the pinned wall behaves as a linear resonator; at higher power the response becomes nonlinear, progressing from localized oscillation to inter-edge relocation and then full escape from the Pt-defined pinning region (Wang et al., 21 Apr 2026). Experimentally, resonant depinning at $$\beta\, \mathbf{m}\times[(\mathbf{u}\cdot\nabla)\mathbf{m}]$$9 and $^3$0 occurs at $^3$1, and at $^3$2 the depinning field begins to decrease above about $^3$3, almost linearly with power. The key point is that the wall mode is spectrally isolated from extended magnons and is directly driven by the antenna field.

Chiral asymmetry produces another depinning mechanism. In CoFeB/Ti/CoFeB trilayers, a uniformly magnetized in-plane CoFeB layer couples to the in-plane moment of homochiral Néel walls in the perpendicular layer, lowering the barrier for one wall type and raising it for the other (Jong et al., 2020). For the $^3$4 state, the down-up wall depins at $^3$5 and the up-down wall at $^3$6; reversing the in-plane layer reverses the asymmetry. The effect can vary the depinning field by up to $^3$7 and decreases monotonically with Ti thickness, consistent with orange-peel magnetostatic coupling rather than oscillatory RKKY.

Other drives act through different torque channels. Pure diffusive spin currents in a nonlocal spin valve reduce the depinning field of a transverse wall with an efficiency of $^3$8, more than an order of magnitude larger than conventional current-induced domain-wall motion in Permalloy, because the absorbed spin current exerts a strong interfacial surface torque where the wall is pinned (Ilgaz et al., 2010). In cylindrical nanowires, an unusually low current density can depin a transverse wall through a rotationally symmetric barrier because the wall is free to rotate around the wire axis; for a $^3$9 barrier the paper reports a current of about $\phi=0$0, corresponding to $\phi=0$1 (Franchin et al., 2011). In VCMA-controlled MTJs, depinning occurs when a pulsed electric field shifts the wall close enough to the Néel-to-Bloch-like transition that the coupled $\phi=0$2 dynamics carries the wall beyond a pinning length $\phi=0$3; with $\phi=0$4, pulse duration determines whether the wall returns to the trap or depins (Upadhyaya et al., 2013). In AFM nanoribbons, the same logic reappears in inertial form: weak damping increases oscillatory overshoot of the wall coordinate and thereby reduces the staggered depinning field (Chen et al., 2019).

4. Scaling, creep, and universality

A major branch of domain wall depinning analysis treats the wall as an elastic interface in quenched disorder. For field-driven ultrathin ferromagnets, a unified creep-and-depinning description uses

$\phi=0$5

with $\phi=0$6, together with the depinning law

$\phi=0$7

and thermal rounding

$\phi=0$8

(Pardo et al., 2016). Using the scaled variables $\phi=0$9 and $q$0, the data collapse onto an empirical universal function $q$1, with $q$2 and $q$3. The same representation collapses Pt/Co/Pt, Au/Co/Au, and CoFeB data.

The subthreshold regime itself can contain additional structure. In [Co/Ni]-based multilayers, the standard creep law underestimates the velocity close to the depinning field because a thermally activated event is followed by a deterministic relaxation whose size grows as $q$4 (Caballero et al., 2017). The corrected model introduces a field-dependent prefactor proportional to the event area, regularized by a cutoff $q$5, and thereby partitions the subthreshold regime into classical creep, an excess-velocity regime, and a saturated-relaxation regime. This permits extraction of $q$6, $q$7, and $q$8 from data below threshold alone when the upward deviation is present.

Microscopic models show where the elastic picture succeeds and where it fails. In the 2D random-field Ising model with dipole-dipole interaction, the depinning threshold for $q$9 is $\psi$0, with $\psi$1, $\psi$2, and $\psi$3, and the subthreshold motion follows a non-Arrhenius activated law,

$\psi$4

where $\psi$5 (Xi et al., 2015). At $\psi$6, by contrast, the activation is Arrhenius-like and the dynamics falls in a different universality class. In the isotropic Ginzburg–Landau study “free of the elastic approximation,” the threshold obeys $\psi$7 for weak random-bond disorder, but overhangs proliferate above a crossover scale $\psi$8 with $\psi$9, so the large-scale geometry crosses over from qEW-like behavior to invasion-percolation-depinning-like behavior (Kolton et al., 2023). In scalar-field simulations of PMA films, the depinning field for uniform disorder of strength $\phi$0 is $\phi$1 at $\phi$2, and the depinning field increases with the mean grain size of a Voronoi tessellation (Caballero et al., 2018).

A separate but related result is that damping itself can renormalize the observed threshold. In chiral PMA systems with DMI, static simulations give a damping-independent field $\phi$3, but dynamic simulations yield a strongly damping-dependent $\phi$4, with $\phi$5 at $\phi$6 because the coupled $\phi$7 dynamics overshoots finite barriers (Moretti et al., 2017). This directly contradicts the conventional assumption that depinning fields are damping independent.

5. Control parameters, trade-offs, and device-oriented optimization

Across the studies, control parameters recur in a relatively small set: notch size and aspect ratio, wire width, angular geometry, grain size, spacer thickness, damping, DMI, drive frequency or pulse width, and the depth and width of the local pinning potential. The reported effects are summarized below.

Control parameter	Reported effect on depinning	Representative sources
Notch size $\phi$8 in Permalloy	$\phi$9 decreases as notch size increases; strong dependence for $H-\beta\chi J$0; crossover near $H-\beta\chi J$1	(Kurniawan et al., 2015)
Segmented-corner field angle	Global minimum near $H-\beta\chi J$2; local minimum near $H-\beta\chi J$3; singularities at $H-\beta\chi J$4 and $H-\beta\chi J$5	(Heinze et al., 2016)
Microwave power in TmIG	Resonant depinning begins near $H-\beta\chi J$6 at $H-\beta\chi J$7; depinning field decreases almost linearly with power	(Wang et al., 21 Apr 2026)
Ti spacer thickness in CoFeB/Ti/CoFeB	Chirality-dependent depinning asymmetry decreases monotonically with increasing Ti thickness and nearly vanishes for Ti(4 nm)	(Jong et al., 2020)
Voronoi grain size	Depinning field increases with mean grain size	(Caballero et al., 2018)
Notch width $H-\beta\chi J$8 in SOT synapses	Wider notches provide better thermal stability–depinning-current trade-off, but increase mean and standard deviation of depinning time	(Kaur et al., 25 Jan 2025)

The device literature makes these trade-offs explicit. In SOT-driven artificial synapses based on triangular notches, the notch-induced pinning potential is fit as

$H-\beta\chi J$9

where $X$0 is the well depth and $X$1 the width parameter (Kaur et al., 25 Jan 2025). Increasing notch depth $X$2 increases $X$3 and $X$4. Increasing notch width parameter $X$5 broadens the well, lowers $X$6, and does not degrade thermal stability significantly, so wider notches provide the better thermal stability–depinning-current trade-off. However, larger $X$7 also increases both the mean and the standard deviation of depinning times at $X$8 K. For exact one-notch-per-pulse programming, the paper proposes the criterion

$X$9

and states that if the pinning strength is too low, typically $J_d$00, no pulse-width optimization exists and programming becomes random. It also notes that a target around $J_d$01 is desirable for retention.

Other applications exploit geometry to rectify motion or encode logic. In triangular antidot arrays, flat-wall depinning obeys

$J_d$02

so $J_d$03 always and flat walls exhibit direct ratchet behavior, while kink propagation can rectify in the opposite direction depending on $J_d$04, generating crossed-ratchet effects (Marconi et al., 2010). In nanowire–nanoparticle gates, strong A/B magnetic configurations produce depinning fields near $J_d$05–$J_d$06 in ideal simulations and about $J_d$07 experimentally, whereas weak C/D configurations have calculated thresholds near $J_d$08–$J_d$09, below the wall nucleation field and therefore experimentally unobservable during reversal (Mironov et al., 2011). This suggests that the practical “depinning threshold” of a device can be masked by a larger nucleation threshold.

6. Limitations, misconceptions, and scope

A persistent misconception is that depinning is determined entirely by static barrier compensation. Several studies contradict this directly. In chiral PMA systems, the dynamic depinning field depends on damping because internal wall dynamics and finite barrier size matter (Moretti et al., 2017). In AFM nanoribbons, weakly damped oscillation of the wall coordinate reduces the required field (Chen et al., 2019). In notched Permalloy wires, the threshold is tied to a specific structural conversion pathway rather than only to a scalar pinning barrier (Kurniawan et al., 2015). A plausible implication is that any depinning analysis that suppresses internal wall degrees of freedom may overestimate thresholds in systems where the wall can oscillate, rotate, or change topology.

A second misconception is that simple geometric projection laws are sufficient whenever the structure is patterned. The segmented-corner sensor study shows otherwise: the depinning field depends on both the observation geometry and the initialized wall position, not on a single tangential field component, and at $J_d$10 the process becomes nucleation-limited rather than propagation-limited (Heinze et al., 2016). Likewise, the isotropic scalar-field study beyond the elastic approximation shows that the directed-interface description is only preasymptotically valid below a disorder-dependent crossover length, after which overhangs and orientational symmetry restoration invalidate a purely single-valued elastic wall picture (Kolton et al., 2023).

The literature also has model-specific limits. The notched Permalloy study includes only adiabatic spin-transfer torque, neglects non-adiabatic torque, thermal fluctuations, external field, and defects beyond the designed notch, and examines only two wire widths and a fixed $J_d$11 thickness (Kurniawan et al., 2015). The electric-field-driven MTJ study is at zero temperature and uses a truncated parabolic pinning potential for depinning (Upadhyaya et al., 2013). The ferrimagnetic microwave study does not provide a compact analytical depinning equation in the main text and models the Pt-induced pinning phenomenologically through a local PMA reduction (Wang et al., 21 Apr 2026). The synapse study relies on a rigid-wall 1D model calibrated against micromagnetics and treats thermal stability through finite-temperature micromagnetic statistics rather than a closed-form escape law (Kaur et al., 25 Jan 2025).

Taken together, these studies define domain wall depinning analysis as a multiscale program. At one end are universal relations for creep, thermal rounding, and depinning exponents; at the other are device-specific thresholds controlled by wall topology, chirality, notch geometry, anisotropy engineering, damping, and pulse protocol. The common technical lesson is that successful depinning analysis must resolve both the structure of the pinning landscape and the internal dynamical pathway by which the wall escapes it.