Dzyaloshinskii–Moriya Interaction (DMI) Overview

• DMI is an antisymmetric exchange interaction arising from spin–orbit coupling and broken inversion symmetry that stabilizes chiral spin textures.
• It is quantified through methods such as domain wall dynamics, Brillouin light scattering, and direct imaging to extract both magnitude and chirality.
• DMI underpins the formation of skyrmions, Néel-type domain walls, and spin spirals, proving critical for designing advanced magnetic devices.

The Dzyaloshinskii–Moriya Interaction (DMI) is an antisymmetric exchange interaction intrinsic to magnetic systems where spin–orbit coupling (SOC) coexists with broken inversion symmetry. DMI favors non-collinear canting of neighboring spins with a fixed handedness ("chirality"), fundamentally underpinning chiral magnetic phenomena, including Néel-type domain walls, spin spirals, skyrmions, and hopfions. DMI manifests at interfaces between ferromagnets (FM) and heavy metals (HM), in noncentrosymmetric bulk crystals, at grain boundaries and microstructural defects, and even in graded or compositionally uniform yet locally non-centrosymmetric systems.

1. Theoretical and Microscopic Foundations

At the atomistic level, the DMI Hamiltonian for localized spins $\mathbf{S}_i$ and $\mathbf{S}_j$ is

$$E_{\mathrm{DMI}} = \sum_{\langle ij\rangle} \mathbf{D}_{ij} \cdot (\mathbf{S}_i \times \mathbf{S}_j)$$

where $\mathbf{D}_{ij}$ is the DMI vector set by the local symmetry, SOC, and inversion asymmetry at the bond (Mazurenko et al., 2021, Hajr et al., 2020, Je et al., 2013). In a continuum (micromagnetic) approximation, the DMI energy density takes the generic Lifshitz-invariant form,

$$w_{\mathrm{DMI}} = D\, \mathbf{m} \cdot (\nabla \times \mathbf{m})$$

where $\mathbf{m}(\mathbf{r})$ is the unit magnetization vector field and $D$ is the DMI constant (units of J/m$^2$ for interfacial cases) (Gueneau et al., 21 Jan 2025, Rashid et al., 2023). For an interface with normal $\hat n$, the relevant symmetry allows only DMI vectors $\mathbf{D}_{ij}$ lying in the plane and perpendicular to the bond (Rowan-Robinson et al., 2017).

Microscopically, the canonical model starts from a Hubbard-plus-SOC Hamiltonian. Moriya’s perturbative expansion (third order in hopping and first in SOC) yields (Fert et al., 2023),

$\mathbf{S}_j$0

where $\mathbf{S}_j$1 is the hopping, $\mathbf{S}_j$2 the atomic SOC strength, and $\mathbf{S}_j$3 the Coulomb energy. Breaking inversion symmetry, by the absence of a center at the bond, is essential; Moriya’s rules precisely determine allowed directions of $\mathbf{S}_j$4 based on point-group symmetries. The resulting antisymmetric exchange fixes a chirality for spin canting (Mazurenko et al., 2021, Fert et al., 2023).

In itinerant electron systems and metallic heterostructures, orbital mixing at interfaces enables finite orbital angular momentum at the Fermi surface (especially between perpendicular-orientation $\mathbf{S}_j$5 orbitals), which is then coupled to spin by SOC, ultimately giving rise to DMI (Hajr et al., 2020). The fundamental Keldysh Green’s function formalism offers a compact and general method for computing the DMI tensor for arbitrary tight-binding Hamiltonians (Hajr et al., 2020).

2. Interfacial DMI in Heterostructures and Thin Films

Interfacial DMI (i-DMI) arises in ultrathin films combining a FM and a HM, where the interface breaks inversion symmetry and the HM provides strong SOC. The DMI constant $\mathbf{S}_j$6 at the interface scales with the strength of SOC and the degree of inversion breaking:

$\mathbf{S}_j$7

where $\mathbf{S}_j$8 is the bond direction and $\mathbf{S}_j$9 the interface normal (Kuepferling et al., 2020). Interfacial DMI is responsible for stabilizing chiral Néel walls and nanoskyrmions in technologically relevant stacks such as Pt/Co, W/CoFeB, Ir/Fe, and Pd/Fe.

The continuum energy density for such bilayers is often expressed as

$$E_{\mathrm{DMI}} = \sum_{\langle ij\rangle} \mathbf{D}_{ij} \cdot (\mathbf{S}_i \times \mathbf{S}_j)$$0

which for a pure $$E_{\mathrm{DMI}} = \sum_{\langle ij\rangle} \mathbf{D}_{ij} \cdot (\mathbf{S}_i \times \mathbf{S}_j)$$1-variation simplifies to $$E_{\mathrm{DMI}} = \sum_{\langle ij\rangle} \mathbf{D}_{ij} \cdot (\mathbf{S}_i \times \mathbf{S}_j)$$2 (Mulkers et al., 2017, Jaiswal et al., 2017). The sign and magnitude of $$E_{\mathrm{DMI}} = \sum_{\langle ij\rangle} \mathbf{D}_{ij} \cdot (\mathbf{S}_i \times \mathbf{S}_j)$$3 are highly sensitive to local atomic structure, interface oxidation, degree of intermixing, and alloy composition (Zimmermann et al., 2018, Kim et al., 2017, Rowan-Robinson et al., 2017). Ab initio computations and experimental studies have shown that DMI remains robust against modest interfacial intermixing but can be tuned (often reduced) by "dusting" the HM interface layer with other elements (especially 4d transition metals), while Au has minimal impact (Zimmermann et al., 2018).

Notably, in Ta/FeCoB/TaOx trilayers, an inversion of DMI chirality with increasing FM thickness (crossover at 5–6 ML) has been observed and rationalized by structural relaxation and orbital-filling-induced sign inversion of the dominant SOC matrix elements (Gueneau et al., 21 Jan 2025). Control of interfacial distance via strain or surface acoustic waves offers a novel route to dynamically tune both magnitude and sign of DMI (Gueneau et al., 21 Jan 2025).

3. Bulk, Gradient-Induced, and Defect DMI

Beyond interfaces and noncentrosymmetric crystals, DMI also emerges in other nontrivial contexts:

• Bulk DMI in Noncentrosymmetric Crystals: Chiral magnets such as B20 (FeGe, MnSi) or polar systems support an intrinsic, symmetry-determined DMI throughout the crystal (Sinaga et al., 2024, Fert et al., 2023). Bulk DMI supports mesoscale chiral phases (helices, skyrmion lattices).
• Gradient-Induced DMI (g-DMI): In films with a compositional gradient of heavy-element atoms along the film normal, a bulk-like, additive DMI arises. Its magnitude and sign can be engineered via the gradient slope, and it scales linearly with thickness (Liang et al., 2022).
• Defect-Induced DMI: Microstructural defects (grain boundaries, dislocations) act as local inversion-breaking centers, engendering chiral exchange even in globally centrosymmetric crystals. The net effect, measurable via polarized SANS, can be substantial (e.g., $$E_{\mathrm{DMI}} = \sum_{\langle ij\rangle} \mathbf{D}_{ij} \cdot (\mathbf{S}_i \times \mathbf{S}_j)$$4 mJ/m$$E_{\mathrm{DMI}} = \sum_{\langle ij\rangle} \mathbf{D}_{ij} \cdot (\mathbf{S}_i \times \mathbf{S}_j)$$5 in Tb) and generalizes DMI to defect-rich polycrystalline systems (Michels et al., 2018).

Furthermore, "hidden" DMI of substantial magnitude has been observed in compositionally uniform, polycrystalline CoPt, ascribed to short-range order or strain gradients, challenging the notion that DMI is exclusive to interfaces or globally noncentrosymmetric crystals (Zhu et al., 2022).

4. Experimental Quantification and Signatures

DMI may be quantified by multiple orthogonal schemes, each probing different magnetic excitations or static configurations, with broad mutual consistency when cross-checked (Kim et al., 2018, Mulkers et al., 2017):

• Domain Wall-Based Methods: The field or current-dependent velocity of DMI-stabilized chiral Néel DWs under in-plane fields yields the DMI-induced "stopping" field $$E_{\mathrm{DMI}} = \sum_{\langle ij\rangle} \mathbf{D}_{ij} \cdot (\mathbf{S}_i \times \mathbf{S}_j)$$6, convertible to $$E_{\mathrm{DMI}} = \sum_{\langle ij\rangle} \mathbf{D}_{ij} \cdot (\mathbf{S}_i \times \mathbf{S}_j)$$7 via

$$E_{\mathrm{DMI}} = \sum_{\langle ij\rangle} \mathbf{D}_{ij} \cdot (\mathbf{S}_i \times \mathbf{S}_j)$$8

(Mulkers et al., 2017, Jaiswal et al., 2017, Kim et al., 2016).

• Spin Wave Spectroscopy: Nonreciprocal propagation of Damon–Eshbach spin waves leads to a frequency shift $$E_{\mathrm{DMI}} = \sum_{\langle ij\rangle} \mathbf{D}_{ij} \cdot (\mathbf{S}_i \times \mathbf{S}_j)$$9, measurable by Brillouin light scattering (BLS) in unpatterned films (Kuepferling et al., 2020, Kim et al., 2018).
• Equilibrium Stripe-Domain and Magnetic Imaging: The equilibrium width of labyrinth or stripe domains depends on DMI-modified domain wall energy $\mathbf{D}_{ij}$0 (Je et al., 2013, Jaiswal et al., 2017).
• Spin-Orbit Torque-Based Measurements: The field dependence of SOT-driven switching events in Hall bars and patterned devices (Kuepferling et al., 2020).
• Direct Structure Imaging: SP-STM, SPLEEM, Lorentz TEM, and NV-center magnetometry can resolve domain wall type and extract $\mathbf{D}_{ij}$1 by modeling the wall profile (Kuepferling et al., 2020).
• Polarized Small-Angle Neutron Scattering (SANS): Detection of DMI-induced chiral terms in the diffuse scattering cross section enables global measurement in bulk, defect-rich, or nanoparticle-assembled systems (Sinaga et al., 2024, Michels et al., 2018).

Quantitative agreement between DW-based and SW-based methods validates the basic phenomenological micromagnetic model of i-DMI across dynamical and static phenomena (Kim et al., 2018). BLS is especially effective for determining both magnitude and sign of $\mathbf{D}_{ij}$2 without device patterning.

For high-precision device design or fundamental studies, cross-validation by at least two orthogonal approaches is recommended.

5. Impact on Chiral Spin Textures and Device Physics

DMI is the essential interaction that stabilizes a range of topologically nontrivial spin textures:

• Néel-type Domain Walls: Chiral walls (fixed sense, such as right- or left-handed) are stabilized for $\mathbf{D}_{ij}$3 exceeding a threshold set by the wall anisotropy and exchange (Je et al., 2013, Gueneau et al., 21 Jan 2025). For $\mathbf{D}_{ij}$4 or $\mathbf{D}_{ij}$5, the favored chirality is set by the sign.
• Skyrmions: Nanoscale skyrmions emerge when $\mathbf{D}_{ij}$6 competes with exchange and anisotropy; their radius $\mathbf{D}_{ij}$7 and threshold for spontaneous formation occurs for $\mathbf{D}_{ij}$8 (Kim et al., 2017, Mulkers et al., 2017, Sinaga et al., 2024).
• Spin Spirals and Helices: In bulk chiral magnets, the DMI sets the helix period $\mathbf{D}_{ij}$9 (Sinaga et al., 2024).
• Confined Chiral Structures: Patterned or spatially modulated DMI can confine DWs, skyrmions, or cycloids, enabling programmable domain and magnonic architectures (Mulkers et al., 2017).
• Topologically Nontrivial 3D Structures: Hopfions and skyrmion tubes are stabilized by DMI in concert with confinement and magnetic anisotropy (Sinaga et al., 2024).

In materials and device development, DMI-driven texturing underpins racetrack memories, SOT-MRAM, skyrmion logic, magnonic waveguides, and is being extended to antiferromagnetic and synthetic multilayer systems (Fert et al., 2023, Liang et al., 2022).

Chirality-inversion phenomena, i.e., sign changes for the DMI as a function of distance or FM thickness, have been detected and explained in atomic-scale detail, with profound consequences for the stability and dynamics of engineered chiral spin textures (Gueneau et al., 21 Jan 2025, Zakeri et al., 2023). The additive character of gradient- or defect-induced DMI further enables scalable chiral-spin functionalities in macroscopically isotropic or even defective media.

6. Novel Mechanisms, Tuning Strategies, and Future Directions

Recent developments highlight advanced mechanisms and potential for engineered control:

• FM Thickness Tuning: The DMI chirality can be reversed by varying FM thickness in HM/FM/oxide trilayers, corresponding to a transition from interface-dominated to decoupled-interface regimes. This is mechanistically linked to hybridization-driven orbital occupation changes and structural relaxation at interfaces (Gueneau et al., 21 Jan 2025).
• Strain and Acoustic Wave Modulation: Strain applied epitaxially or via surface acoustic waves tunes both the magnitude and chirality of DMI dynamically, allowing real-time manipulation of chiral textures (Gueneau et al., 21 Jan 2025).
• Gradient-Induced and Defect-Engineered DMI: Deliberate doping or strain gradients produce robust, thickness-scalable bulk DMI with tunable handedness, while control of essential defect densities offers another axis for DMI engineering (Liang et al., 2022, Michels et al., 2018).
• Interfacial DMI in Complex Magnetic Orders: Antiferromagnetic and noncollinear systems (e.g., stacked-Kagome) exhibit DMI-induced emergent anisotropies, enabling control of multistate switching and SOT-driven phenomena (Yamane et al., 16 Feb 2025).

At the theoretical level, continued development of first-principles-accurate computational protocols (e.g., Keldysh GF, generalized Bloch theorem, magnetic force theorem, Berry-phase approaches) allows quantitative prediction of DMI in realistic heterostructures and low-dimensional materials (Hajr et al., 2020, Mazurenko et al., 2021, Kim et al., 2017).

7. Tables: Experimental Methods for Quantifying DMI

Method	Magnetic Quantity Probed	DMI Range/Accuracy
DW velocity/depining	Chiral domain wall dynamics	$$w_{\mathrm{DMI}} = D\, \mathbf{m} \cdot (\nabla \times \mathbf{m})$$0 pJ/m, sign+mag
Brillouin light scat.	Non-reciprocal SW frequency	$$w_{\mathrm{DMI}} = D\, \mathbf{m} \cdot (\nabla \times \mathbf{m})$$1 pJ/m, high accuracy
Stripe/annihilation	Equilibrium domain pattern	$$w_{\mathrm{DMI}} = D\, \mathbf{m} \cdot (\nabla \times \mathbf{m})$$2 pJ/m, magnitude only
SOT-based	Field/switch current in Hall bar	$$w_{\mathrm{DMI}} = D\, \mathbf{m} \cdot (\nabla \times \mathbf{m})$$3 pJ/m, sign+mag
SANS (polycrystals)	Chiral spin correlations	Bulk and defect DMI
DW internal ang., SPLEEM/NV	DW profile imaging	Direct, sign+mag, atomic to nm

These methods yield complementary measurements, and cross-validation is standard for high-fidelity DMI parameter extraction in device prototyping and fundamental research (Kuepferling et al., 2020, Kim et al., 2018).

---

References: (Gueneau et al., 21 Jan 2025, Hajr et al., 2020, Kim et al., 2017, Mulkers et al., 2017, Jaiswal et al., 2017, Perrey et al., 2013, Kuepferling et al., 2020, Zimmermann et al., 2018, Kim et al., 2018, Fert et al., 2023, Sinaga et al., 2024, Zhu et al., 2022, Liang et al., 2022, Yamane et al., 16 Feb 2025, Zakeri et al., 2023, Mazurenko et al., 2021), and others as specified inline.