Field-Free Magnetization Reversal

• Field-free magnetization reversal is the process of switching magnetic states without an external field by using built-in symmetry-breaking mechanisms.
• Multiple approaches—including spin Hall torques, exchange bias, orbital currents, and optical stimuli—are used to deterministically lift state degeneracy.
• Reversal dynamics involve domain nucleation, chirped drives, and parametric resonance, underscoring challenges in achieving uniform and scalable device performance.

Field-free magnetization reversal denotes the reversal of a magnetic order parameter without an externally applied magnetic field. In the contemporary spintronics literature, the term spans deterministic switching of perpendicular ferromagnetic magnetization, reversal of ferrimagnetic moments, reorientation of a perpendicular Néel vector, and even reprogramming of exchange bias by electrical or optical stimuli. The central technical problem is symmetry selection: in perpendicularly magnetized systems, the two opposite out-of-plane states are degenerate unless a built-in symmetry-breaking element or a dynamically selective drive is present. Recent work has therefore developed several routes, including spin Hall and spin-orbit torques combined with exchange bias or interlayer exchange coupling, orbital-current and orbit-transfer mechanisms, resonant and chirped drives, multiferroic domain-wall decoupling, and single-shot laser control of exchange bias (Brink et al., 2015, Liu et al., 2018, Ye et al., 2022, Riel et al., 2024).

1. Symmetry selection as the defining requirement

A recurring result across the literature is that field-free reversal is not simply the absence of an external field; it is the replacement of that field by an intrinsic or dynamically generated symmetry-breaking mechanism. In perpendicular magnetic anisotropy devices driven by spin-orbit torque, conventional switching typically requires an in-plane field or an unconventional configuration to lift the $+z/-z$ degeneracy. This limitation is stated explicitly for both spin-Hall-based perpendicular switching and related SOT geometries (Brink et al., 2015, Ye et al., 2022).

Several distinct symmetry-breaking strategies have been demonstrated or proposed. In Pt/Co/IrMn, an in-plane exchange bias established by field cooling substitutes for the external in-plane field and enables deterministic spin-Hall switching of a perpendicularly magnetized Co layer (Brink et al., 2015). In Co/Ru/Co/Ir/Co/Ta, a $1.35\ \mathrm{nm}$ Ir layer simultaneously generates SOT through the spin Hall effect and mediates antiferromagnetic interlayer exchange coupling to an in-plane synthetic antiferromagnet, producing an effective in-plane exchange field of approximately $20\ \mathrm{mT}$ (Liu et al., 2018). In Co/PtGd/Mo and Mo/CoGd, lateral symmetry breaking associated with a pronounced planar Hall effect yields a finite out-of-plane damping-like torque, while orbital Hall currents amplify the in-plane damping-like torque needed for deterministic reversal (Bekele et al., 2024, Bekele et al., 9 Jun 2025). In an AFM/Insulator/Heavy Metal trilayer, the required selection is produced by the coexistence of a uniform damping-like torque from one current and a staggered field-like torque from an orthogonal current, permitting reversible $180^\circ$ switching of a perpendicular Néel vector (Xu et al., 2022).

This multiplicity of routes establishes a general principle: field-free reversal is fundamentally a symmetry-engineering problem rather than a single mechanism class. A second consequence is that “field-free” does not imply “bias-free.” Exchange bias, interlayer exchange, orbital polarization, planar-Hall-induced $p_z$, and staggered torques all act as internal selectors of the final state.

2. Spin-Hall and exchange-coupled electrical switching

The most direct field-free electrical route in the recent literature combines spin Hall torque with an internally generated in-plane bias. In Ta(1)/Pt(3)/Co(0.7)/Pt(0.3)/IrMn(6)/TaOx(1.5), annealing at $225^\circ\mathrm{C}$ in a $2\ \mathrm{T}$ in-plane field for $30\ \mathrm{min}$ followed by field cooling establishes an in-plane exchange bias in IrMn, while Pt provides the spin Hall current source. Deterministic switching is observed with no external magnetic field during the measurement, with current pulses of approximately $8\times 10^{11}\ \mathrm{A/m^2}$ for $50\ \mu\mathrm{s}$ (Brink et al., 2015). The same study emphasizes that the reversal is often gradual and partial rather than perfectly binary, and attributes this to the polycrystalline IrMn microstructure, which produces local variations in exchange-bias magnitude and direction.

A related but structurally distinct implementation uses interlayer exchange coupling rather than exchange bias. In Substrate/Co(2)/Ru(0.85)/Co(2)/Ir(1.35)/Co(1.2)/Ta(2), the top $1.35\ \mathrm{nm}$0 Co layer has strong perpendicular magnetic anisotropy, and the Ir spacer serves a dual role: it generates SOT via the spin Hall effect and mediates antiferromagnetic interlayer exchange coupling to the lower Co/Ru/Co synthetic antiferromagnet (Liu et al., 2018). Hall measurements show that conventional SOT switching appears for $1.35\ \mathrm{nm}$1, the switching hysteresis collapses at $1.35\ \mathrm{nm}$2 because the external field cancels the interlayer exchange field, and robust deterministic switching reappears at $1.35\ \mathrm{nm}$3 due to the intrinsic exchange field. Control samples without the bottom synthetic antiferromagnet, or with thicker Ir at $1.35\ \mathrm{nm}$4, do not show field-free switching.

These two systems illustrate two different forms of internal in-plane bias. Exchange bias in Pt/Co/IrMn is interfacial and tied to uncompensated antiferromagnetic moments and grain structure (Brink et al., 2015). Interlayer exchange in Ir-mediated stacks is spacer-thickness-dependent and can exceed the DMI effective field at the Ir/Co interface, reported as approximately $1.35\ \mathrm{nm}$5, thereby restoring domain-wall expansion in all directions (Liu et al., 2018). Both approaches are directly motivated by perpendicular MRAM architectures, because they retain the write-path separation of SOT while avoiding a macroscopic applied field.

3. Orbital and parametric current-driven mechanisms

Field-free reversal is no longer confined to spin-current physics in the narrow sense. A major extension is orbit-transfer torque in WTe$1.35\ \mathrm{nm}$6/Fe$1.35\ \mathrm{nm}$7GeTe$1.35\ \mathrm{nm}$8. In this heterostructure, few-layer WTe$1.35\ \mathrm{nm}$9 possesses a nonzero Berry curvature dipole, and an in-plane current along the crystal $20\ \mathrm{mT}$0-axis generates a perpendicular polarization of orbital magnetic moments, described as $20\ \mathrm{mT}$1 (Ye et al., 2022). The adjacent Fe$20\ \mathrm{mT}$2GeTe$20\ \mathrm{mT}$3 is the perpendicular ferromagnet. Deterministic switching is observed when the current is applied along the $20\ \mathrm{mT}$4-axis, whereas current along the $20\ \mathrm{mT}$5-axis produces no deterministic switching and instead yields multi-domain states. For device A at $20\ \mathrm{mT}$6, the critical current density is $20\ \mathrm{mT}$7; for device C it is $20\ \mathrm{mT}$8, and switching remains robust up to $20\ \mathrm{mT}$9 (Ye et al., 2022).

A second orbital route uses the orbital Hall effect as an auxiliary or primary source of angular momentum. In Ta(0.5)/Mo($180^\circ$0)/Pt$180^\circ$1Gd$180^\circ$2(2)/Co(0.8)/Pt(1.5), the addition of a Mo underlayer enhances the in-plane SOT efficiency $180^\circ$3 by $180^\circ$4–$180^\circ$5 relative to the no-Mo case, while the planar Hall effect supplies a finite $180^\circ$6 and thus a moderate out-of-plane damping-like torque (Bekele et al., 2024). The paper reports robust field-free switching in $180^\circ$7-diameter pillar-shaped devices, with switching efficiency close to $180^\circ$8 for both $180^\circ$9 and $p_z$0 Mo. In a related Mo/CoGd device, the CoGd layer acts both as an orbital-to-spin converter and as the ferrimagnetic switching layer itself, eliminating the need for an additional conversion layer. The reported critical current density for deterministic field-free switching is as low as $p_z$1 at $p_z$2, and the mechanism is associated with localized orbital Hall currents in Mo, in-plane symmetry breaking at the Mo/CoGd interface, and a substantial unconventional $p_z$3-polarized damping-like torque (Bekele et al., 9 Jun 2025).

A third route, conceptually different from both SHE and OHE, is the parametric mechanism in a FeB nanomagnet on W. There the key ingredient is a current-induced in-plane magnetic field $p_z$4 produced by spin accumulation at the nanomagnet interfaces. In a $p_z$5 FeB layer on a $p_z$6 tungsten Hall bar, $p_z$7 reaches about $p_z$8 at a current density of $p_z$9, increases linearly with current, and is measured with $225^\circ\mathrm{C}$0 precision (Zayets, 2021). Although much smaller than the $225^\circ\mathrm{C}$1 anisotropy field, it is sufficient for parametric resonance. The analytical torque is written as

$225^\circ\mathrm{C}$2

with $225^\circ\mathrm{C}$3 (Zayets, 2021). This places resonance-assisted DC switching alongside spin and orbital torque as a third electrical paradigm.

4. Reversal dynamics: domain walls, chirped drives, dipolar engineering, and phase-mediated pathways

Field-free switching is often discussed as if it were a macrospin instability, but several studies show that the microscopic path can be domain-wall-mediated or strongly nonequilibrium. In the Ir-mediated SOT stack, Kerr microscopy reveals that reversal begins with nucleation of small reversed domains, often at device edges, and subsequent $225^\circ\mathrm{C}$4 current pulses expand those domains by current-driven domain-wall motion. Final reversal can remain incomplete because of stubborn domains, which the study attributes to local anisotropy variations (Liu et al., 2018). Micromagnetic modeling further indicates isotropic wall expansion and a dependence of wall velocity on the orientation of the wall core magnetization relative to the wall tangent.

A different nonequilibrium route is chirped current-pulse SOT. Numerical work on a perpendicularly magnetized nanodevice shows that a circularly polarized chirped current pulse,

$225^\circ\mathrm{C}$5

can use both field-like and damping-like SOT components to achieve field-free ultrafast reversal without symmetry-breaking means (Liu et al., 2024). The reported minimal current density for the ferromagnetic case is $225^\circ\mathrm{C}$6, compared with conventional SOT requiring $225^\circ\mathrm{C}$7 for similar reversal rates. The same framework gives $225^\circ\mathrm{C}$8 for a perpendicular synthetic antiferromagnet and $225^\circ\mathrm{C}$9 for a synthetic ferrimagnet, both with pulse duration of about $2\ \mathrm{T}$0, and stochastic LLG simulations indicate robustness at $2\ \mathrm{T}$1 (Liu et al., 2024).

Related resonant control appears in down-chirp microwave reversal of a single-domain magnetic nanoparticle. Numerical simulations based on the LLG equation show subnanosecond reversal with a circularly polarized down-chirp microwave field of amplitude $2\ \mathrm{T}$2, initial frequency $2\ \mathrm{T}$3, and chirp rate up to $2\ \mathrm{T}$4, with switching time down to $2\ \mathrm{T}$5 (Islam et al., 2018). A constant-frequency microwave field at $2\ \mathrm{T}$6 requires $2\ \mathrm{T}$7 for comparable reversal speed. The reported mechanism is that the pulse acts as an energy source before the barrier crossing and as an energy sink afterward.

Dipolar interaction engineering provides a distinct static-energy-landscape route. In a two-body Stoner-particle system with synchronized dynamics, the modified Stoner–Wohlfarth result

$2\ \mathrm{T}$8

implies zero-field reversal at the critical dipolar interaction strength $2\ \mathrm{T}$9 in the perpendicular configuration (Sun et al., 2010). For cobalt nanoparticles with $30\ \mathrm{min}$0, $30\ \mathrm{min}$1, and $30\ \mathrm{min}$2, the paper gives $30\ \mathrm{min}$3, a critical spacing $30\ \mathrm{min}$4, a reduced switching field of about $30\ \mathrm{min}$5 compared with $30\ \mathrm{min}$6, and reversal times of $30\ \mathrm{min}$7–$30\ \mathrm{min}$8 (Sun et al., 2010).

A conceptually related, though not itself presented as a field-free protocol, is magnetization reversal through an antiferromagnetic state in Zn-doped Fe$30\ \mathrm{min}$9Mo$8\times 10^{11}\ \mathrm{A/m^2}$0O$8\times 10^{11}\ \mathrm{A/m^2}$1. There the coercive point is accompanied by atomic-scale compensation via transient recovery of the antiferromagnetic phase rather than by mesoscopic domain cancellation, as shown by a sharp polarization peak and reappearance of the AFM magnon mode in THz spectroscopy (Ghara et al., 2022). This suggests that phase-mediated compensation may become relevant for future field-free control schemes that exploit competing free-energy minima instead of only torque balance.

5. Electric-field, optical, and antiferromagnetic implementations

Electrical current is not the only route to field-free reversal. In a $8\times 10^{11}\ \mathrm{A/m^2}$2 Permalloy film on single-crystal hexagonal LuMnO$8\times 10^{11}\ \mathrm{A/m^2}$3, electric pulses such as $8\times 10^{11}\ \mathrm{A/m^2}$4, approximately $8\times 10^{11}\ \mathrm{A/m^2}$5, $8\times 10^{11}\ \mathrm{A/m^2}$6, triangular profile, applied near coercivity induce abrupt isothermal magnetization reversal and switch the sign of the exchange bias (Skumryev et al., 2010). LuMnO$8\times 10^{11}\ \mathrm{A/m^2}$7 is antiferromagnetic below $8\times 10^{11}\ \mathrm{A/m^2}$8, ferroelectric up to $8\times 10^{11}\ \mathrm{A/m^2}$9, and not ferroelastic, which the study uses to rule out strain-mediated switching. The proposed mechanism is electric-field-driven decoupling of ferroelectric and antiferromagnetic domain walls: fast ferroelectric domain-wall motion leaves the clamped antiferromagnetic walls behind, transiently removing pinning of the Permalloy and allowing reversal, after which reclamping restores the exchange bias with opposite sign (Skumryev et al., 2010).

Optical reprogramming of exchange bias adds an ultrafast route. In Ta(4)/Pt(4)/Co(0.6)/Gd(5.5)/Co(1)/[Pt(1.25)/Co(0.6)]$50\ \mu\mathrm{s}$0/IrMn($50\ \mu\mathrm{s}$1)/Ta(5), a single $50\ \mu\mathrm{s}$2 laser pulse reverses the Co/Gd synthetic ferrimagnet through helicity-independent all-optical switching and then imprints the new orientation into IrMn, yielding field-free $50\ \mu\mathrm{s}$3 reorientation of perpendicular exchange bias (Riel et al., 2024). The Co/Gd reversal occurs within about $50\ \mu\mathrm{s}$4 in the layered three-temperature model, while exchange-bias setting in granular IrMn is described by an Arrhenius law over intermediate to slow times up to milliseconds. The work reports up to $50\ \mu\mathrm{s}$5 consecutive switches without degradation in exchange-bias magnitude and preservation of unidirectionality, expressed as $50\ \mu\mathrm{s}$6 (Riel et al., 2024).

Antiferromagnets can also be switched directly. In the proposed AFM/Insulator/Heavy Metal trilayer, a current in the heavy metal generates a uniform damping-like torque through the spin Hall effect, while an orthogonal current in the antiferromagnet generates a staggered field-like torque because of locally broken inversion symmetry (Xu et al., 2022). The combined torques permit reversible field-free switching of an antiferromagnet with perpendicular Néel vector. The reported threshold current density is $50\ \mu\mathrm{s}$7, switching completes within approximately $50\ \mu\mathrm{s}$8, and the two sublattice moments become temporarily noncollinear, with a maximum angle difference of about $50\ \mu\mathrm{s}$9 (Xu et al., 2022). This is a qualitatively different form of field-free reversal because the relevant order parameter is the Néel vector rather than a net ferromagnetic moment.

6. Device implications, limitations, and recurrent misconceptions

For memory applications, the immediate attraction of field-free reversal is architectural. SOT-based approaches avoid passing high-density current through a tunnel barrier, and the Ir-based interlayer-exchange design explicitly argues that the dual functionality of Ir permits future build-up of magnetoresistive stacks for memory and logic applications (Liu et al., 2018). The exchange-bias Pt/Co/IrMn work similarly frames built-in symmetry breaking as a way to avoid local field coils or global in-plane fields in perpendicular MRAM (Brink et al., 2015). Orbit-transfer torque and orbital-Hall-enabled switching are likewise presented as routes toward low-power, scalable, reliable perpendicular switching without external fields (Ye et al., 2022, Bekele et al., 9 Jun 2025).

A recurrent misconception is that “field-free” implies spatially uniform coherent reversal. The experimental literature shows otherwise. In Ir-mediated SOT switching, reversal proceeds by domain nucleation and expansion rather than by uniform rotation (Liu et al., 2018). In Pt/Co/IrMn, gradual and incomplete switching arises from a statistical distribution of local exchange-bias magnitude and direction at a polycrystalline ferromagnet/antiferromagnet interface (Brink et al., 2015). These observations indicate that internal symmetry breaking can be highly nonuniform at device scale.

A second misconception is that field-free reversal is necessarily a spin-current problem. The WTe$1.35\ \mathrm{nm}$00/Fe$1.35\ \mathrm{nm}$01GeTe$1.35\ \mathrm{nm}$02 work identifies orbital magnetic moments generated by a Berry curvature dipole as the active polarization source (Ye et al., 2022). The Mo/PtGd/Co and Mo/CoGd studies use orbital Hall currents and orbital-to-spin conversion to amplify or directly generate the effective torques needed for deterministic reversal (Bekele et al., 2024, Bekele et al., 9 Jun 2025). Parametric torque in FeB introduces yet another language in which a current-induced in-plane magnetic field, rather than a conventional Slonczewski spin torque alone, is the essential resonant drive (Zayets, 2021).

The maturity of the different routes also varies. Pt/Co/IrMn, Ir-mediated Co-based stacks, WTe$1.35\ \mathrm{nm}$03/Fe$1.35\ \mathrm{nm}$04GeTe$1.35\ \mathrm{nm}$05, Mo/PtGd/Co, Mo/CoGd, Py/LuMnO$1.35\ \mathrm{nm}$06, and Co/Gd/IrMn are experimental demonstrations (Brink et al., 2015, Liu et al., 2018, Ye et al., 2022, Bekele et al., 2024, Bekele et al., 9 Jun 2025, Skumryev et al., 2010, Riel et al., 2024). By contrast, chirped-current SOT in ferromagnetic, SAF, and SFi nanodevices is a numerical demonstration, the orthogonal-current AFM trilayer is a proposal supported by macrospin modeling, the down-chirp microwave protocol is numerical, and the dipolar two-Stoner-particle scheme is theoretical (Liu et al., 2024, Xu et al., 2022, Islam et al., 2018, Sun et al., 2010). This suggests that “field-free magnetization reversal” is best treated as a family of mechanisms with different experimental readiness, not as a single settled technology.

The present literature therefore supports a broad but technically coherent definition. Field-free reversal comprises any switching process in which deterministic state selection is achieved without an external magnetic field, whether the selector is exchange bias, interlayer exchange coupling, orbital polarization, planar-Hall-induced symmetry breaking, parametric resonance, chirped energy transfer, multiferroic domain-wall dynamics, optical imprinting into an antiferromagnet, or direct antiferromagnetic torque engineering. Across these routes, the governing issues remain the same: how symmetry is broken, how angular momentum or free energy is redistributed during reversal, and how reproducibly the final state can be selected in realistic, nonuniform materials systems.