FeGaB/LiNbO3 Magnetoacoustic Devices

• FeGaB/LiNbO3 magnetoacoustic devices are hybrid structures merging piezoelectric LiNbO3 with magnetostrictive FeGaB to enable SAW-driven spin-wave excitation and nonreciprocal behavior.
• The platform leverages precise anisotropy engineering, interdigital transducer patterning, and dipolar interactions to control magnetoelastic coupling and directional acoustic attenuation.
• Nonlinear spin-wave dynamics and localized acoustic scattering in these devices offer potential for high-sensitivity magnetic sensing and spatially resolved imaging.

FeGaB/LiNbO$_3$ magnetoacoustic devices are hybrid piezoelectric–ferromagnetic heterostructures in which surface acoustic waves (SAWs) launched on LiNbO$_3$ couple to the magnetization dynamics of FeGaB through magnetoelastic interaction. In the best studied FeGaB implementations, LiNbO$_3$ functions as a long-decay-length, efficient SAW transport medium, while FeGaB provides low coercive field, self-biased magnetic behavior, strong magnetoelastic coupling, nonreciprocal SAW behavior, and low damping. The resulting platform supports SAW-driven spin-wave excitation, directional acoustic attenuation, and nonlinear magnon processes, and has been developed for isolator-like acoustic transport as well as for spatially resolved studies of spin-acoustic dynamics (Shah et al., 2020, Beaver et al., 14 Jul 2025).

1. Device platform and structural realizations

A canonical FeGaB/LiNbO$_3$ realization is a two-port SAW delay line on single-crystal $y$-cut LiNbO$_3$, with split-finger interdigital transducers defining Rayleigh-wave propagation along the $z$-axis and a FeGaB/Al$_2$O$_3$/FeGaB multilayer deposited in the acoustic path. In the isolator geometry, each FeGaB layer is $20$ nm thick, the Al$_3$0O$_3$1 spacer is $_3$2 nm, and the patterned magnetic region is $_3$3 wide and $_3$4 long. The IDTs comprise $_3$5 finger pairs with minimum electrode spacing $_3$6, the IDT spacing is $_3$7 mm, and the Al electrode thickness is $_3$8 nm. The same platform is measured mainly at odd harmonics, especially $_3$9 MHz, although $_3$0 MHz and $_3$1 MHz are also reported (Shah et al., 2020).

A second FeGaB/LiNbO$_3$2 implementation, used for spatially resolved nonlinear-dynamics studies, is a hybrid magnetoacoustic heterostructure comprising a LiNbO$_3$3 SAW substrate, patterned IDTs, and a FeGaB bilayer

$_3$4

This device is driven at the $_3$5th harmonic, $_3$6 MHz, a mode previously identified as having exceptionally strong magnetoelastic response (Beaver et al., 14 Jul 2025).

Across LiNbO$_3$7-based magnetoacoustic devices more generally, substrate cut and SAW propagation direction are decisive structural parameters. Related LiNbO$_3$8 SAW devices based on Ni employ Y-cut Z-propagation and $_3$9 Y-cut configurations, while direct-written magnetic micro-stripes on standard Y-cut Z-propagation LiNbO$_3$0 demonstrate how patterned magnetic elements can be selectively addressed within a SAW aperture. These systems do not use FeGaB, but they establish the broader LiNbO$_3$1 design space within which FeGaB/LiNbO$_3$2 devices are situated (Sasaki et al., 2016, Küß et al., 2022, Yu et al., 3 Sep 2025).

2. Magnetoelastic transduction and resonance physics

The central mechanism is acoustically driven magnetization dynamics. A SAW propagating on LiNbO$_3$3 generates time-varying strain at the surface; in the magnetostrictive FeGaB film this strain acts as an effective drive field, transferring acoustic energy into spin excitations when the magnetic system satisfies the relevant resonance and wavevector conditions. In the FeGaB bilayer isolator, the coupling is expressed through the magnetoelastic energy density

$_3$4

with $_3$5 the strain tensor and $_3$6 the local magnetization direction. The key physical point is that the SAW carries a definite phase relation among strain components, so the overlap between the acoustic mode and the magnetic eigenmode depends on propagation direction, field orientation, and magnetic free energy (Shah et al., 2020).

In the Ni/LiNbO$_3$7 formulation, which is described as directly analogous in device physics and modeling, the lowest-order magnetoelastic coupling is written as

$_3$8

For a SAW propagating along $_3$9, the relevant strain components include $y$0, $y$1, and $y$2, with the shear term phase shifted relative to the longitudinal components. Reversing $y$3 reverses the sign of the shear-related term, thereby changing the interference between coupling channels and, in turn, the absorption and phase velocity. This provides the basic directional magnetoelastic logic later exploited in FeGaB bilayers (Sasaki et al., 2016).

A standard in-plane Rayleigh-wave selection rule is that magnetoelastic coupling is maximized when

$y$4

However, LiNbO$y$5-based hybrid devices show that this textbook condition is not sufficient by itself, because anisotropy and dipolar fields shift the equilibrium magnetization angle away from the naive geometrical expectation. This is especially relevant to FeGaB/LiNbO$y$6, where the same ferromagnetic-thin-film-on-piezoelectric-substrate architecture applies but with stronger magnetoelastic response than Ni (Yu et al., 3 Sep 2025).

The broader LiNbO$y$7 SAW literature also establishes that Rayleigh waves provide a nontrivial vertical shear component $y$8. In Y-cut Z-propagation LiNbO$y$9, this component enables magnetoelastic excitation of forward volume spin waves even when the static field is nearly or exactly along the film normal. That result was demonstrated for CoFe-based micro-stripes rather than FeGaB, but it identifies a coupling channel that is directly relevant for FeGaB/LiNbO$_3$0 devices using comparable Rayleigh-wave geometries (Küß et al., 2022).

3. Magnetic anisotropy, dipolar terms, and free-energy engineering

Magnetoacoustic response in LiNbO$_3$1 hybrids is not determined solely by magnetostriction. In Ni/LiNbO$_3$2 resonators, magneto-optical Kerr effect measurements revealed both a twofold symmetry component and a fourfold symmetry component, demonstrating biaxial anisotropy in addition to uniaxial terms. The coercivity angular dependence was fitted as

$_3$3

and the easy-axis–hard-axis separations were reported as $_3$4 and $_3$5 for two propagation geometries rather than the $_3$6 expected for simple uniaxial anisotropy. The authors argue that this anisotropy is likely substrate-induced, arising from interfacial strain related to the crystallographic orientation of LiNbO$_3$7 (Yu et al., 3 Sep 2025).

That result is directly relevant to FeGaB/LiNbO$_3$8. The same extraction states that FeGaB is magnetostrictive and typically more anisotropy-sensitive than Ni, so the same design lesson applies with potentially stronger consequences. In this view, the LiNbO$_3$9 cut and SAW direction can imprint anisotropy in the magnetic film, alter the free-energy landscape, shift the equilibrium magnetization angle, and thereby move the absorption maxima away from $z$0. A plausible implication is that FeGaB/LiNbO$z$1 optimization requires explicit control of anisotropy symmetry rather than postulating a single easy axis (Yu et al., 3 Sep 2025).

Anisotropy alone is not enough. The Ni/LiNbO$z$2 analysis explicitly adds a dipolar interaction term in addition to uniaxial anisotropy, biaxial anisotropy, and exchange. Calculations including biaxial anisotropy plus dipolar interactions reproduce the angular SAW absorption data, whereas a uniaxial-only model fails. For FeGaB/LiNbO$z$3, the same design rule is stated explicitly: choose the LiNbO$z$4 cut and SAW direction deliberately, characterize whether the film acquires uniaxial, biaxial, or mixed anisotropy, include dipolar terms in the model, and tune the film anisotropy so the equilibrium magnetization angle matches the SAW magnetoelastic selection rule (Yu et al., 3 Sep 2025).

The FeGaB bilayer isolator adds a second, more structured source of magnetic asymmetry: dynamic dipolar coupling between the two FeGaB layers across the ultrathin Al$z$5O$z$6 spacer. In the bilayer spin-wave theory, the mutual cross-demagnetization tensor is non-symmetric under wavevector inversion,

$z$7

and is written as

$z$8

This sign$z$9 dependence is the necessary condition for spin-wave nonreciprocity in the FeGaB bilayer spectrum (Shah et al., 2020).

4. Nonreciprocal SAW propagation and isolator behavior

The most distinctive FeGaB/LiNbO$_2$0 functionality demonstrated so far is giant nonreciprocity. In the FeGaB/Al$_2$1O$_2$2/FeGaB bilayer on $_2$3-cut LiNbO$_2$4, the SAW dispersion crosses the spin-wave dispersion for one propagation direction but remains separated for the opposite direction. Under that condition, the backward-propagating SAW resonates strongly with the magnetic mode and is heavily damped, whereas the forward-propagating SAW is off-resonant and transmits with much lower loss. At $_2$5 MHz and with the growth field at $_2$6 to the SAW axis, the measured isolation is $_2$7 dB. The maximum observed transmission is about $_2$8 dB, minimum transmission reaches $_2$9 dB at $_3$0 Oe and $_3$1 for forward propagation and at $_3$2 Oe and $_3$3 for reverse propagation, and the minimum transmission is about $_3$4 dB below the insertion-loss floor, corresponding to roughly $_3$5 power absorption at resonance (Shah et al., 2020).

This nonreciprocal state is field and angle selective. A central qualitative statement is that resonant absorption occurs only in the directions perpendicular to the growth field, not along it. When the growth field is at $_3$6 to the SAW axis, the interaction appears strongly in quadrants II and IV and is absent in quadrants I and III. At $_3$7 relative to the SAW axis, the nonreciprocity largely vanishes, and only weak absorption remains because the magnetic state is canted and some residual magnetoelastic coupling survives (Shah et al., 2020).

Earlier Ni/LiNbO$_3$8 studies established the same symmetry logic in a simpler single-layer setting. There, absorption and phase velocity depend on the sign of wave vector, and the symmetry relation

$_3$9

captures the observed reciprocity constraint. Reversing the in-plane magnetic field direction by $20$0 reverses the nonreciprocal response pattern, and the magnitude of the effect depends strongly on ferromagnetic film shape: a rectangular Ni film shows clear nonreciprocity, whereas a circular Ni film shows nearly negligible nonreciprocity. This shape dependence was attributed to different balances of magnetoelastic coupling channels, and it provides a direct precedent for geometry-sensitive directional control in FeGaB/LiNbO$20$1 devices (Sasaki et al., 2016).

The principal limitation of the FeGaB isolator is not the absence of nonreciprocity but the presence of high insertion loss. The baseline transmission is roughly $20$2 dB even away from resonance, which the source attributes to IDT mismatch, air loading, higher-harmonic operation, and coupling to thermally excited elastic waves. Practical deployment therefore requires simultaneous control of two quantities that are partly independent: directional magnetic absorption and non-magnetic acoustic loss (Shah et al., 2020).

5. Nonlinear spin-wave dynamics and NV-based microscopy

FeGaB/LiNbO$20$3 devices do not operate only in a linear, spatially uniform regime. In NV-center imaging of a FeGaB/LiNbO$20$4 heterostructure driven at $20$5 MHz, the acoustically induced magnetic response was found to be highly heterogeneous across micron length scales and strongly nonlinear in drive power. The experiment used a dense ensemble of NV centers implanted about $20$6 nm below the diamond surface, with the diamond placed directly on the FeGaB/LiNbO$20$7 device so that the NVs sensed local magnetic noise from the driven film. Electrical transmission shows an acoustic linewidth of about $20$8 MHz, whereas the NV-detected ODMR linewidth is about $20$9 MHz, indicating that the NVs predominantly detect a narrower nonlinear thresholded process rather than the full linear acoustic resonance (Beaver et al., 14 Jul 2025).

Time-domain relaxometry revealed threshold behavior. Under microwave drive, the NV spin population decays as

$_3$00

and the spin-lattice relaxation rate increases by more than two orders of magnitude when the microwave drive power is tripled. The acoustic drive therefore increases the spectral density of magnetic noise at frequencies above the NV ground-state transition, $_3$01 GHz, even though the SAW drive itself is at $_3$02 MHz. The source interprets this as evidence for a nonlinear frequency-doubling or multistep magnon-scattering process rather than direct resonant excitation of the NV ground state (Beaver et al., 14 Jul 2025).

The observed inhomogeneity is dynamic rather than static. Spatial maps show structures on micron to $_3$03 length scales; the patterns persist over repeated field sweeps, yet MOKE imaging confirms that they are not static magnetic domains. The FeGaB region is described as essentially uniformly magnetized and having very low coercivity, $_3$04 G. The heterogeneity is therefore attributed to local wavevector conditions, likely set by acoustic scattering from receiver electrodes, topographical defects, or other acoustic inhomogeneities in the LiNbO$_3$05/film structure (Beaver et al., 14 Jul 2025).

The proposed nonlinear mechanism is a two-step cascade: magnon confluence produces a higher-frequency mode near

$_3$06

and above threshold that mode undergoes first-order Suhl instability, generating broadband incoherent magnetic noise in the NV-sensitive band. Micromagnetic simulations using the FeGaB bilayer dispersion from MuMax3 support the plausibility of this pathway and show strong angle sensitivity, with the nonlinear process peaking when the second wavevector satisfies approximately

$_3$07

The same study explicitly rules out several simpler interpretations: direct ground-state NV driving, coherent excited-state driving, simple second-harmonic generation, and a thermal explanation requiring $_3$08 K. It also rejects a spin-chemical-potential explanation because the extracted value would require $_3$09, which is described as unphysical under the experimental conditions (Beaver et al., 14 Jul 2025).

6. Sensing, spatial selectivity, and transferable design rules

Two related literatures clarify the broader operating envelope of FeGaB/LiNbO$_3$10 magnetoacoustics. First, a non-LiNbO$_3$11 but directly relevant Love-wave platform on ST-cut quartz with a SiO$_3$12 guiding layer and polycrystalline Fe$_3$13Ga$_3$14 shows how strong shear-horizontal polarization can amplify field sensitivity near coercivity. In that system, the best-performing device used a $_3$15 nm FeGa film and, at $_3$16, exhibited a sharp frequency peak near $_3$17 mT, a maximum frequency shift of about $_3$18 kHz, and a magnetic sensitivity of $_3$19 Hz/nT. The response polarity changes with field angle, being upward at $_3$20 and downward at $_3$21, and the acoustic frequency of approximately $_3$22 MHz is far below the FMR frequency of FeGa, so the response is quasi-static rather than resonant-FMR-driven. Although this is not a FeGaB/LiNbO$_3$23 device, it demonstrates that coercive-field operation and shear-horizontal surface waves can create a distinct sensitivity regime for magnetostrictive SAW sensors (Aguilera et al., 14 Mar 2025).

Second, Y-cut Z-propagation LiNbO$_3$24 devices with tapered interdigital transducers show how SAW frequency can be mapped onto spatial position, allowing micron-scale addressing of adjacent magnetic elements with a single transducer pair. In that geometry, the local resonance frequency varies along the aperture according to

$_3$25

and the beam-width estimate

$_3$26

gives $_3$27. The same work demonstrates that the Rayleigh-wave vertical shear strain $_3$28 can excite forward volume spin waves and can generate nonreciprocity in geometries where the effective driving field acquires helicity. These results were obtained in CoFe-based stripes, not FeGaB, but they define a directly relevant LiNbO$_3$29 transduction and addressing strategy (Küß et al., 2022).

Taken together with the anisotropy study on Ni/LiNbO$_3$30, these related platforms imply a coherent design program for FeGaB/LiNbO$_3$31. The recurring message is that optimization requires anisotropy engineering rather than simply maximizing magnetostriction; that realistic modeling must include exchange and dipolar terms; that propagation direction, substrate cut, and field angle jointly determine reciprocity and absorption; and that microscopic acoustic scattering can become a dominant noise source even when the ferromagnet appears magnetically uniform. A plausible implication is that future FeGaB/LiNbO$_3$32 devices will be most effective when acoustic mode structure, anisotropy symmetry, and scattering environment are co-designed rather than treated as separable subsystems (Yu et al., 3 Sep 2025, Beaver et al., 14 Jul 2025, Küß et al., 2022).