Plasmon-Enhanced Inverse Faraday Effect

• The paper explores how plasmonic nanostructures amplify local optical spin density to induce a stationary, quasi-static magnetization via the inverse Faraday effect.
• It details electromagnetic mechanisms such as field amplification, phase retardation, and rectification that enable sub-picosecond ultrafast magnetization switching.
• Experimental and simulation data show that optimized plasmonic geometries can yield effective magnetic fields ranging from mT to Tesla levels with enhanced energy efficiency.

to=python code to=python code to=python code to=python code to=python code Plasmon-enhanced inverse Faraday effect denotes the generation of a stationary magnetization or effective quasi-static magnetic field by optical excitation whose local spin density is amplified, phase-structured, or spatially confined by plasmonic modes. In its common phenomenological form, the optomagnetic response is written as $$B_{\mathrm{IFE}}(\mathbf r)=\beta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$, $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$, or, for an isotropic magneto-optical medium, $$\mathbf H_{\mathrm{eff}}(\mathbf r)=-i\varepsilon_0\beta[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$. In metals, equivalent current-based descriptions attribute the effect to optically driven circulating electron motion, orbital magnetization, or rectified drift currents that generate a stationary near magnetic field. Plasmonic nanostructures enhance the effect because localized surface plasmon resonances and surface plasmon polaritons concentrate electromagnetic fields, increase longitudinal field components and phase retardation, and create hotspots of $\operatorname{Im}(\mathbf E\times \mathbf E^*)$ at the nanoscale (Hareau et al., 30 Jun 2025).

1. Electromagnetic basis and quasi-static rectification

The inverse Faraday effect is a nonlinear light–matter interaction in which a medium acquires a stationary magnetization when driven by light with nonzero optical spin. A convenient gauge-invariant quantity for design and interpretation is the optical spin density

$$\mathbf S=\frac{\varepsilon_0}{4\omega}\operatorname{Im}(\mathbf E\times \mathbf E^*)+\frac{\mu_0}{4\omega}\operatorname{Im}(\mathbf H\times \mathbf H^*),$$

with the electric term often dominating in metal near fields. In isotropic and centrosymmetric descriptions the induced magnetization is collinear with $i(\mathbf E\times \mathbf E^*)$, so its sign follows the local helicity. For circular polarization $E=E_0(\hat x\pm i\hat y)/\sqrt{2}$, one has $\operatorname{Im}(E_xE_y^*)=\pm |E_0|^2/2$; for general elliptical polarization, $$\operatorname{Im}(E_xE_y^*)=|E_x||E_y|\sin\Delta\phi$$ (Hareau et al., 30 Jun 2025).

Plasmonic systems are distinctive because they need not rely on a free-space circularly polarized input to generate nonzero $i(\mathbf E\times \mathbf E^*)$. In the surface-plasmon-polariton case analyzed for propagation along $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$0, the relevant invariant is

$$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$1

so a transverse-magnetic mode with longitudinal field $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$2 and normal field $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$3 carries the spin density that drives the IFE even when the launched light is linearly $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$4-polarized. For a single-interface SPP, the amplitude ratio is

$$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$5

with $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$6 the dielectric permittivity and $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$7 the metal permittivity. Near the SPP resonance, $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$8, while away from resonance it remains moderate at $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$9, and the sign changes between dielectric and metal (Im et al., 2018).

The induced field is quasi-static because $$\mathbf H_{\mathrm{eff}}(\mathbf r)=-i\varepsilon_0\beta[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$0 is already a time-averaged invariant over the optical cycle for a monochromatic or narrowband field. For a two-frequency excitation,

$$\mathbf H_{\mathrm{eff}}(\mathbf r)=-i\varepsilon_0\beta[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$1

the self-terms are strictly DC, whereas the cross terms oscillate at $$\mathbf H_{\mathrm{eff}}(\mathbf r)=-i\varepsilon_0\beta[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$2. If $$\mathbf H_{\mathrm{eff}}(\mathbf r)=-i\varepsilon_0\beta[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$3 is small compared to the inverse timescale of magnetization response or the pulse-envelope duration, these cross terms act as slowly varying quasi-static bias fields. This rectification mechanism underlies proposals for sub-ps magnetization control by two-frequency pulses in magneto-plasmonic cavities (Im et al., 2018).

2. Enhancement channels and canonical plasmonic platforms

Plasmonic enhancement proceeds through two levers emphasized across the literature: local-field amplification and spin-density hotspots. Near a localized surface plasmon resonance, the enhancement factor $$\mathbf H_{\mathrm{eff}}(\mathbf r)=-i\varepsilon_0\beta[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$4 can reach $$\mathbf H_{\mathrm{eff}}(\mathbf r)=-i\varepsilon_0\beta[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$5–$$\mathbf H_{\mathrm{eff}}(\mathbf r)=-i\varepsilon_0\beta[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$6 depending on geometry and material, and the resulting IFE scales with the local $$\mathbf H_{\mathrm{eff}}(\mathbf r)=-i\varepsilon_0\beta[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$7. Equally important, plasmonic phase retardation and modal interference create large $$\mathbf H_{\mathrm{eff}}(\mathbf r)=-i\varepsilon_0\beta[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$8 even under nominally linear far-field drive, so the near-field spin can exceed the far-field helicity bounds; the review literature describes this regime as “super-circular” light (Hareau et al., 30 Jun 2025).

Different geometries realize this enhancement through different current patterns, field gradients, and symmetry constraints. Localized resonances in discs, rods, dimers, and chiral antennas emphasize electric-field hotspots and near-field helicity engineering; propagating SPPs in planar and MIM geometries emphasize longitudinal-field-driven spin density; hydrodynamic electron liquids and periodically modulated rings emphasize rectified plasmonic vortex currents (Im et al., 2018).

Platform	Key mechanism	Reported outcome
Ag/BIG MIM waveguide–cavity	SPP longitudinal field and detuning-controlled cavity phase	$$\mathbf H_{\mathrm{eff}}(\mathbf r)=-i\varepsilon_0\beta[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$9 T in the gap and $\operatorname{Im}(\mathbf E\times \mathbf E^*)$0 T in the cavity (Im et al., 2018)
Au nanodiscs on Tb$\operatorname{Im}(\mathbf E\times \mathbf E^*)$1Co$\operatorname{Im}(\mathbf E\times \mathbf E^*)$2	LSPR-enhanced near fields and interface-localized optomagnetic field	$\operatorname{Im}(\mathbf E\times \mathbf E^*)$3 about an order of magnitude stronger than in bare TbCo (Parchenko et al., 2023)
Achiral Au nanodisk arrays	CP-driven circulating electron motion modifies damping	reflectance increase by 78%, field concentration increase by 35.7%, $\operatorname{Im}(\mathbf E\times \mathbf E^*)$4 T (Cheng et al., 2022)
Au nanorod under linear polarization	modal phase retardation creates “super circular light”	$\operatorname{Im}(\mathbf E\times \mathbf E^*)$5 mT and about 25 times the sphere reference (Yang et al., 2022)
Inversely designed chiral Au antenna	geometry-sculpted spin density for one-helicity-only response	strong magnetic field of 500 mT only for one helicity (Mou et al., 2023)
Plasmonic crystal ring / 2DEL coupler	rectified twisted plasmons and vortex-state dc current	giant IFE with mT-scale to tens-of-mT-scale fields in disk and solenoid architectures (Aizin et al., 2021, Potashin et al., 2020)

In the nanoparticle limit, semiclassical quantum-hydrodynamic modeling of small Au nanoparticles attributes the induced magnetization to rotating surface currents driven by resonant circular excitation. For radii $\operatorname{Im}(\mathbf E\times \mathbf E^*)$6–$\operatorname{Im}(\mathbf E\times \mathbf E^*)$7 nm, the model yields center fields $\operatorname{Im}(\mathbf E\times \mathbf E^*)$8, $\operatorname{Im}(\mathbf E\times \mathbf E^*)$9, $$\mathbf S=\frac{\varepsilon_0}{4\omega}\operatorname{Im}(\mathbf E\times \mathbf E^*)+\frac{\mu_0}{4\omega}\operatorname{Im}(\mathbf H\times \mathbf H^*),$$0, and $$\mathbf S=\frac{\varepsilon_0}{4\omega}\operatorname{Im}(\mathbf E\times \mathbf E^*)+\frac{\mu_0}{4\omega}\operatorname{Im}(\mathbf H\times \mathbf H^*),$$1 T at $$\mathbf S=\frac{\varepsilon_0}{4\omega}\operatorname{Im}(\mathbf E\times \mathbf E^*)+\frac{\mu_0}{4\omega}\operatorname{Im}(\mathbf H\times \mathbf H^*),$$2, and about $$\mathbf S=\frac{\varepsilon_0}{4\omega}\operatorname{Im}(\mathbf E\times \mathbf E^*)+\frac{\mu_0}{4\omega}\operatorname{Im}(\mathbf H\times \mathbf H^*),$$3atom at $$\mathbf S=\frac{\varepsilon_0}{4\omega}\operatorname{Im}(\mathbf E\times \mathbf E^*)+\frac{\mu_0}{4\omega}\operatorname{Im}(\mathbf H\times \mathbf H^*),$$4 at plasmon resonance (Hurst et al., 2018).

3. Detuning, propagation direction, and polarization engineering

A central result of the waveguide–cavity formulation is that the sign of the quasi-static IFE field can be controlled spectrally. In the side-coupled magneto-optical cavity, the cavity-averaged field obeys

$$\mathbf S=\frac{\varepsilon_0}{4\omega}\operatorname{Im}(\mathbf E\times \mathbf E^*)+\frac{\mu_0}{4\omega}\operatorname{Im}(\mathbf H\times \mathbf H^*),$$5

and the transmission satisfies $$\mathbf S=\frac{\varepsilon_0}{4\omega}\operatorname{Im}(\mathbf E\times \mathbf E^*)+\frac{\mu_0}{4\omega}\operatorname{Im}(\mathbf H\times \mathbf H^*),$$6. At exact resonance a standing wave forms, so both the energy flow and the net IFE field vanish inside the cavity; at opposite detunings $$\mathbf S=\frac{\varepsilon_0}{4\omega}\operatorname{Im}(\mathbf E\times \mathbf E^*)+\frac{\mu_0}{4\omega}\operatorname{Im}(\mathbf H\times \mathbf H^*),$$7, the cavity supports equal-magnitude fields with opposite signs. Reversing the SPP propagation direction likewise reverses the sign. In full-field simulations, forward SPPs at one detuning produce $$\mathbf S=\frac{\varepsilon_0}{4\omega}\operatorname{Im}(\mathbf E\times \mathbf E^*)+\frac{\mu_0}{4\omega}\operatorname{Im}(\mathbf H\times \mathbf H^*),$$8 T inside the cavity, whereas a different detuning or backward SPPs yield $$\mathbf S=\frac{\varepsilon_0}{4\omega}\operatorname{Im}(\mathbf E\times \mathbf E^*)+\frac{\mu_0}{4\omega}\operatorname{Im}(\mathbf H\times \mathbf H^*),$$9 T (Im et al., 2018).

Detuning control also appears in localized-plasmon platforms, but there it is coupled to the heating problem. In Au nanodisc arrays on Tb$i(\mathbf E\times \mathbf E^*)$0Co$i(\mathbf E\times \mathbf E^*)$1, finite-difference time-domain calculations showed that at resonance the electric field is strongly concentrated at disc edges and absorption is enhanced in both the plasmonic metal and the adjacent magnetic film, whereas off resonance the electric-field enhancement is reduced but the near-field magnetic distribution becomes strongly concentrated at the Au/TbCo interface. Experimentally, the largest helicity-dependent response $i(\mathbf E\times \mathbf E^*)$2 occurred for excitation detuned to wavelengths larger than the plasmon resonance, realized at fixed $i(\mathbf E\times \mathbf E^*)$3 nm by choosing larger disc radii; $i(\mathbf E\times \mathbf E^*)$4 at $i(\mathbf E\times \mathbf E^*)$5 nm was $i(\mathbf E\times \mathbf E^*)$6 lower than at $i(\mathbf E\times \mathbf E^*)$7 nm (Parchenko et al., 2023).

A common misconception is that only circular or elliptical illumination can drive the IFE. In homogeneous media, ideal linear polarization gives zero net IFE, but plasmonic near fields can invalidate that far-field intuition. A gold nanorod excited by linear polarization at an angle $i(\mathbf E\times \mathbf E^*)$8 supports longitudinal and transverse dipolar modes with a resonance-induced phase retardation, generating local elliptical polarization and nonzero spin density although the incident light remains strictly linear. The net field at the rod center follows $i(\mathbf E\times \mathbf E^*)$9, with maxima at $E=E_0(\hat x\pm i\hat y)/\sqrt{2}$0 and $E=E_0(\hat x\pm i\hat y)/\sqrt{2}$1, zeros at $E=E_0(\hat x\pm i\hat y)/\sqrt{2}$2, $E=E_0(\hat x\pm i\hat y)/\sqrt{2}$3, and $E=E_0(\hat x\pm i\hat y)/\sqrt{2}$4, and sign reversal by rotation of the input polarization (Yang et al., 2022).

Plasmonic geometry can go further and reverse or chiralize the usual helicity rule. In an inversely designed chiral Au antenna at $E=E_0(\hat x\pm i\hat y)/\sqrt{2}$5 nm, the optimized pattern produces a strong magnetic field of 500 mT only for one helicity, while the mirror structure activates the opposite helicity. In the related reversed-IFE design, the near-field spin-density hotspot has sign opposite to the incident helicity, so right circular polarization generates a magnetization opposite to the propagation direction and left circular polarization generates it along the propagation direction; in that study the field reached about 6 mT at the structure center for a $E=E_0(\hat x\pm i\hat y)/\sqrt{2}$6 fs pulse at $E=E_0(\hat x\pm i\hat y)/\sqrt{2}$7, and the normalized spin density reached $E=E_0(\hat x\pm i\hat y)/\sqrt{2}$8 (Mou et al., 2023, Mou et al., 2023).

4. Ultrafast dynamics, switching, and nonlinear self-action

The magnetization dynamics driven by plasmon-enhanced IFE are usually cast in the Landau–Lifshitz–Gilbert framework,

$E=E_0(\hat x\pm i\hat y)/\sqrt{2}$9

with $\operatorname{Im}(E_xE_y^*)=\pm |E_0|^2/2$0. For the BIG cavity in the MIM waveguide, magnetization saturates at $\operatorname{Im}(E_xE_y^*)=\pm |E_0|^2/2$1 T, and an incident SPP mode power of $\operatorname{Im}(E_xE_y^*)=\pm |E_0|^2/2$2 was estimated to generate $\operatorname{Im}(E_xE_y^*)=\pm |E_0|^2/2$3 T in the cavity, sufficient to reach saturation and drive reversal on sub-ps timescales, with $\operatorname{Im}(E_xE_y^*)=\pm |E_0|^2/2$4 fs cited as consistent with ultrafast switching times reported in the literature (Im et al., 2018).

Pump–probe measurements in Au nanodisc/TbCo heterostructures confirmed that plasmonic enhancement can change both amplitude and temporal character. In the Au-nanodisc/TbCo system, $\operatorname{Im}(E_xE_y^*)=\pm |E_0|^2/2$5 rises essentially at time zero, within the pump pulse duration, which is consistent with direct plasmonic electron drift generating an immediate optomagnetic field. In bare TbCo films without Au, the $\operatorname{Im}(E_xE_y^*)=\pm |E_0|^2/2$6 maximum occurs at $\operatorname{Im}(E_xE_y^*)=\pm |E_0|^2/2$7 ps, consistent with an intrinsic IFE mediated by electron–spin coupling after the initial electronic excitation (Parchenko et al., 2023).

The IFE also enters nonlinear plasmonics as a self-action mechanism. For planar SPPs in magnetoplasmonic structures, the longitudinal electric field acts through the IFE as an effective transverse magnetic field, and Lorentz-reciprocity analysis yields the cubic envelope equation

$\operatorname{Im}(E_xE_y^*)=\pm |E_0|^2/2$8

For a ferromagnetic dielectric/metal interface and for a dielectric/hybrid Au–Co–Au interface, analytical expressions were derived for the effective third-order coefficient $\operatorname{Im}(E_xE_y^*)=\pm |E_0|^2/2$9, with an order-of-magnitude estimate of $$\operatorname{Im}(E_xE_y^*)=|E_x||E_y|\sin\Delta\phi$$0 on the order of $$\operatorname{Im}(E_xE_y^*)=|E_x||E_y|\sin\Delta\phi$$1. The same analysis used $$\operatorname{Im}(E_xE_y^*)=|E_x||E_y|\sin\Delta\phi$$2 as a representative IFE susceptibility, compared with $$\operatorname{Im}(E_xE_y^*)=|E_x||E_y|\sin\Delta\phi$$3 near $$\operatorname{Im}(E_xE_y^*)=|E_x||E_y|\sin\Delta\phi$$4–$$\operatorname{Im}(E_xE_y^*)=|E_x||E_y|\sin\Delta\phi$$5 nm (Im et al., 2018).

Hydrodynamic formulations supply an orbital-current counterpart to these field-based descriptions. In a gated two-dimensional electron liquid, circularly polarized radiation and a plasmonic coupler excite twisted plasmonic oscillations whose nonlinear rectification produces a helicity-sensitive dc current and an out-of-plane magnetic moment. In periodic arrays the current takes the resonant form

$$\operatorname{Im}(E_xE_y^*)=|E_x||E_y|\sin\Delta\phi$$6

with the plasmonic and mixed terms producing Fano-like asymmetry. This framework makes the IFE simultaneously a source of optical magnetization and a contactless diagnostic of momentum relaxation and viscosity in electron fluids (Potashin et al., 2020).

5. Quantitative performance, efficiency, and damping control

Reported field magnitudes span from mT to Tesla, depending on geometry, excitation modality, and whether the quoted value is directly inferred, simulated, or defined through an effective-field mapping. In the Ag/BIG MIM geometry, the transverse distribution across the gap at $$\operatorname{Im}(E_xE_y^*)=|E_x||E_y|\sin\Delta\phi$$7 nm reaches $$\operatorname{Im}(E_xE_y^*)=|E_x||E_y|\sin\Delta\phi$$8 T for an incident SPP mode power of $$\operatorname{Im}(E_xE_y^*)=|E_x||E_y|\sin\Delta\phi$$9, with opposite signs at the top and bottom metal interfaces; in the side-coupled cavity, the internal field reaches $i(\mathbf E\times \mathbf E^*)$0 T for appropriate detuning and propagation direction (Im et al., 2018). In the chiral inverse-design antenna, the reported stationary field is 500 mT at $i(\mathbf E\times \mathbf E^*)$1 (Mou et al., 2023). In the linearly excited Au nanorod, the simulated $i(\mathbf E\times \mathbf E^*)$2 at the center is on the order of 200 mT at $i(\mathbf E\times \mathbf E^*)$3 (Yang et al., 2022).

Efficiency estimates are most explicit for the waveguide-integrated switching proposal. Using waveguide lateral width $i(\mathbf E\times \mathbf E^*)$4 and pulse FWHM $i(\mathbf E\times \mathbf E^*)$5 fs, the required energy was estimated as $i(\mathbf E\times \mathbf E^*)$6 fJ/bit for $i(\mathbf E\times \mathbf E^*)$7–$i(\mathbf E\times \mathbf E^*)$8 T IFE fields. The same work contrasts this with traditional free-space IFE in bulk films, which typically requires $i(\mathbf E\times \mathbf E^*)$9 fluence for $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$00 T fields over tens of $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$01m spots, corresponding to $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$02 nJ/bit (Im et al., 2018). A plausible implication is that the dominant gain comes from nanoscale confinement and direct waveguide delivery rather than from a universal increase of the intrinsic magneto-optical coefficient.

Not all studies extract absolute $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$03 directly. In the Au-nanodisc/TbCo experiments, the primary observables were $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$04, $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$05, and transient transmittance; absolute $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$06 or $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$07 values in Tesla or mT were not directly extracted because conversion requires the material-specific $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$08 or $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$09 (Parchenko et al., 2023). By contrast, the achiral Au nanodisk damping study used an external field of $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$10 T and thermometric calibration to infer an effective magnetic field of $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$11 T during circularly polarized CW excitation at $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$12. That same study reported a reversible reflectance increase of 78%, a 35.7% increase in effective field concentration at hotspots, and a damping change from $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$13 meV to $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$14 meV, implying $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$15 (Cheng et al., 2022).

These results show that plasmon-enhanced IFE is not only a means of generating magnetic bias fields but also a route for modifying plasmonic dissipation itself. In the magneto-Drude picture used for achiral Au nanodisks,

$$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$16

so the mode co-rotating with the electron cyclotron motion experiences a reduction in effective damping. This is a distinct operating regime from magnetization switching, but it arises from the same circular electron trajectories and stationary optomagnetic response (Cheng et al., 2022).

6. Limitations, measurement challenges, and unresolved questions

Heating, absorption, and spectral sensitivity recur throughout the field. SPP excitation in Ag entails loss, and the BIG cavity proposal explicitly requires careful pulse-energy management and heat dissipation. Au nanodisc arrays on TbCo were operated at $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$17 to avoid damage, and higher fluences caused nanostructure damage. The design logic that emerges is therefore not simply “drive at resonance”: near-resonance excitation increases both IFE and absorption, whereas off-resonance excitation can reduce demagnetization while sustaining strong chiral near-field distributions (Im et al., 2018, Parchenko et al., 2023).

Fabrication tolerances and calibration are equally consequential. In the MIM waveguide–cavity system, the sign and magnitude of $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$18 depend sensitively on the SPP wavelength and cavity detuning, requiring precise control of cavity dimensions of $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$19 nm thickness and $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$20 nm length, and an MIM gap of $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$21–$$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$22 nm (Im et al., 2018). The chiral and reversed-IFE antennas rely on binary patterns with $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$23 nm pixels and smoothed corners, so any experimental implementation must preserve the phase landscape that generates the target $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$24 hotspots (Mou et al., 2023, Mou et al., 2023).

The most persistent controversy concerns magnitude. A 2025 review identifies a stark, often orders-of-magnitude, mismatch between theoretically predicted and experimentally inferred magnetization values across metallic and plasmonic IFE studies. Contributing factors summarized there include material modeling, dissipation and dephasing, interband and spin–orbit effects, cancellation between macroscopic drift and microscopic orbital terms, thermal versus nonthermal signal contamination, and the fact that many field values are inferred optically rather than measured magnetically (Hareau et al., 30 Jun 2025).

This suggests that progress depends as much on metrology as on geometry design. The same review argues for direct, high-resolution probes of transient magnetic fields at their native length and time scales, identifying ultrafast electron microscopy and Lorentz TEM, nitrogen-vacancy center magnetometry, and time-resolved near-field magneto-optic imaging as promising routes. In parallel, platform-specific calibration of $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$25, $$M_{\mathrm{IFE}}(\mathbf r)=\eta(\omega)\operatorname{Im}[\mathbf E(\mathbf r)\times \mathbf E^*(\mathbf r)]$$26, or equivalent current-based response functions remains necessary if relative helicity-dependent observables are to be converted into absolute nanoscale magnetic fields with reproducible accuracy (Hareau et al., 30 Jun 2025).