Introduction

The internet of things (IoT) is a network of interconnected devices and objects that collect, communicate and exchange information autonomously. It has permeated numerous sectors of modern life, from smart homes and wearable gadgets to industrial automation and health-care monitoring. The proliferation of IoT devices necessitates solutions that address key challenges, including compact form factors, energy efficiency, seamless communication and adaptability to diverse environments1. Magnetoelectric (ME) materials, composites and devices offer unique functionalities and capabilities that can revolutionize the landscape of IoT technologies.

ME systems are composed of engineered materials and composites with strong direct magnetoelectric coupling and converse magnetoelectric coupling. The origin of ME materials traces back to the discovery and exploration of single-phase ME materials from the late nineteenth century through the mid-twentieth century. In the 1970s, scientists at the Philips Research Laboratories studied ME composites based on in situ grown eutectic piezoelectric and ferrite materials, setting the groundwork for subsequent developments in ME composites and devices2,3. In 2001, the demonstration of the strong ME coupling of piezoelectric–magnetostrictive laminate composites ignited an exponential growth of research and development on ME materials and devices4,5, encompassing applications such as ME-based magnetic field sensors, memory and logic components, and power, radiofrequency (RF) and microwave applications, antennas and filters2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32 (Fig. 1).

Fig. 1: Timeline of key contributions to ME materials and devices.
figure 1

On the timeline, the seminal works are colour coded to exemplify whether they are material advancements (yellow), direct magnetoelectric (ME) coupling (blue), converse ME coupling (red) or both direct and converse ME coupling (purple)2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31. Applications of direct and converse ME coupling mechanisms are summarized in the Venn diagram. VLF, very low frequency. WPT, wireless power transfer.

Full size image

ME sensors can detect magnetic fields on the order of fT at room temperature19 and enable ultra-compact ME mechanical antennas for mobile bi-directional very-low-frequency (VLF) communications29,30. Furthermore, ME devices show enhanced efficiency for wireless power transfer (WPT) over state-of-the-art resonant coils and have the potential to facilitate simultaneous wireless information and power transfer26. Leveraging distinct acoustic resonance modes, simultaneous magnetic field sensing, VLF, RF and microwave communications, and WPT have been experimentally demonstrated26.

One of the prominent advantages of ME devices, especially for mechanical antennas, is their ultra-compact size. Because ME antennas operate at their electromechanical resonance, their size is comparable to the mechanical wavelength rather than the electromagnetic wavelength. Being that the mechanical wavelength is 10−5 the magnitude of the electromagnetic wavelength, ultra-miniaturized RF components are permitted32. This reduction in size is beneficial for IoT systems wherein space constraints dictate device designs. Together with ME WPT, ME energy harvesters can power IoT nodes using ambient mechanical vibrations, and magnetic and electric fields, reducing dependence on traditional power sources and prolonging the operational lifespan of battery-powered devices. From improved efficiency and compact sizes to magnetic field sensing and data processing, ME technologies can reshape the IoT landscape, propelling its evolution towards a streamlined, interconnected, and intelligent ecosystem.

In this Review, we provide a foundational understanding of the cutting-edge progress in piezoelectric and magnetic materials for ME composites and discuss their application in memory and logic devices33, such as magnetoelectric spin–orbit (MESO)-coupled logic devices34. Then we present a range of ME microelectromechanical and nanoelectromechanical systems (M/NEMS) relevant to IoT applications22,26,31,32, such as magnetic field sensing for wearable medical devices26, antennas designed for both ultra-high-frequency (UHF, 0.3–3 GHz)22,31 and very-low-frequency (VLF, 3–30 kHz) bands29,30, and wireless power transfer26. Furthermore, we highlight ME-tunable RF components35,36 and introduce pioneering devices such as acoustic-driven ferromagnetic magnetic resonance (ADFMR)37 for nonreciprocal RF devices25,28, quantum transduction and control38,39,40, and VLF antenna arrays29,41 for underground and underwater communications. Finally, we present a prospective analysis of the highlights and challenges of the use of ME M/NEMS systems in IoT.

ME materials and composites

Strain-mediated ME coupling is a dominant mechanism to describe ME coupling42. Strain-mediated ME laminates composed of piezoelectric and magnetostrictive materials demonstrate much higher ME coupling strength (ME coupling on the order of hundreds of mV cm−1 Oe−1) than single-phase multiferroic materials (two or more ferroic properties, such as ferromagnetism, antiferromagnetism, ferrimagnetism, ferroelectricity, antiferroelectricity, ferroelasticity and  ferrotoroidicity), such as bismuth ferrite (BiFeO3, BFO has an ME coupling of ~0.064 mV cm−1 Oe−1)32,43. For bulk laminate composites, the theoretical ME voltage coefficient \({\alpha }_{{\rm{ME}}}^{L-L}\) for the longitudinal–longitudinal (L–L) mode is given by equation (1):

$${\alpha }_{{\rm{ME}}}^{L-L}=\frac{n{d}_{33,m}\,{g}_{33,p}}{n{s}_{33}^{E}\left(1-{{k}_{33}}^{2}\right)+\left(1-n\right){s}_{33}^{H}}$$
(1)

where \({s}_{33}^{E}\) and \({s}_{33}^{H}\) are the elastic compliances of the piezoelectric and magnetostrictive layers, respectively; k33 is the electromechanical coupling coefficient of the piezoelectric layer; n is the ratio of magnetic layer thickness to the total thickness of the ME laminate; \({d}_{33,m}\) and \({g}_{33,p}\) are the longitudinal piezomagnetic and piezoelectric coefficients, respectively13.

The assumption here is that the interfacial coupling coefficient is equal to 1. The interface between the magnetic and piezoelectric phases is perfectly bonded and enables strain transfer without appreciable loss. Additionally, it is assumed that there is no dielectric, piezoelectric, and mechanical loss in the composite. The ME voltage coefficient depends on the electromechanical and magnetomechanical properties of the piezoelectric and ferromagnetic layers.

The magnetic material in the ME composites should have high magnetostriction (>70 ppm), high piezomagnetic coefficient (>10 ppm Oe−1), near unity magnetomechanical coupling factor, and low loss (ferromagnetic resonance linewidth <20 Oe in X-band)14. High ME coupling has been achieved in magnetic alloys, such as (Fe80Ga20)88B12 (ref. 14), or ferrite films16 grown on single crystal ferroelectric substrates, namely Pb(Zn1/3Nb2/3)O3–PbTiO3 (PZN-PT) or Pb(Mg1/3Nb2/3)O3–PbTiO3 (PMN-PT), enabling high mechanical resonance quality factor (QMR), low-frequency devices17,44,45,46,47. However, the large electric fields (8 kV cm−1 translating to 400 V for 0.5-mm-thick ferroelectric slab) required to tune the magnetization are not feasible for most working devices44. Additionally, the single-crystal ferroelectrics are not easily integrated onto Si substrates owing to back-end-of-line complementary metal-oxide-semiconductor (CMOS) processing incompatibility (that is crystallization temperature higher than 450 °C), and the requirement of lattice matching for epitaxy. Thin film ME heterostructures have been integrated with CMOS-compatible processes on Si using piezoelectric films, such as AlN, ZnO, Al0.73Sc0.27N, coupled to magnetostrictive alloy films, such as (Fe80Ga20)88B12 or (Fe90Co10)78Si12B10, to fabricate ME devices for both communication and sensing applications22,48,49,50.

For strain-mediated ME composites, it is critical to select piezoelectric materials with a high piezoelectric coefficient. Although perovskite ferroelectrics can exhibit high piezoelectric coefficients51, they are not compatible with CMOS technology owing to the requirement of epitaxial growth. Therefore, AlN and Sc-doped AlN are considered the premier candidates for use in ME composites51,52. Fabrication methods have also been developed to integrate ferroelectric lithium niobate, LiNbO3, on Si wafers via ion slicing and film transfer approach53,54, thus enabling LiNbO3 deployment in future ME-based IoT devices.

Doping galfenol (Fe80Ga20) films with boron or carbon causes a crystallographic phase transition from the body-centred cubic phase to an amorphous structure, leading to excellent soft magnetic properties (<2 Oe coercive field), high magnetostriction (78 ppm), near unity magnetomechanical coupling factor (0.84), and low ferromagnetic resonance linewidth in X-band (<20 Oe linewidth) in the body-centred cubic-amorphous phase boundary region14,36,55,56, making Fe1−xGax alloys the most promising magnetic materials for ME devices14,55,56,57. Novel growth approaches utilizing molecular beam epitaxial growth of Fe1−xGax films on PMN-PT have expanded the possible composition range for the disordered body-centred cubic (A2) phase space, achieving an ultra-high magnetostriction at a Ga concentration of x = 30% with a diminishing stiffness, thus limiting their applications58. Underlayers such as NiFe, Cu, Cr, or Ti seed layer (<5 nm) before magnetic film deposition were demonstrated to promote soft magnetic properties in Fe-Ga alloys59.

Aside from the magnetomechanical properties and static magnetic behaviour, magnetic loss is a key property, particularly for RF ME NEMS. Magnetic loss can be understood by characterizing the ferromagnetic resonance linewidth and/or the Gilbert damping parameter60. Yttrium iron garnet (Y3Fe5O12, YIG) films are the pinnacle for low magnetic losses with an intrinsic magnetic damping parameter as low as approximately 3 × 10–5, with the lowest reported magnetic damping parameter for the thin film regime at approximately 5.2 × 10–5 (refs. 61,62). This low magnetic damping enables long spin wave decay lengths of more than 100 μm (refs. 63,64). However, a key challenge for YIG integration is the lattice-matching requirement for epitaxy, specifically requiring growth on a Gd3Ga5O12 (GGG) substrate, which is not back-end-of-line CMOS compatible65,66. Although the GGG substrate allows for heteroepitaxial growth of YIG films, the GGG substrate can serve as a source of contamination in sputtered YIG films67,68. Although liquid-phase epitaxy grown YIG films have shown minimal to no contamination from the GGG substrate, liquid-phase epitaxy is not as back end of line compatible as sputtering69. In 2022, a buffer layer of diamagnetic Y3Sc2.5Al2.5O12 thin film on GGG enabled the growth of YIG films with off-axis sputtering with suppressed Gd-contamination to enable decent Gilbert damping at few-kelvin (K) temperatures70. Despite recent work suppressing Gd-contamination from the GGG substrate, there is a need for more investigation into the integration of YIG with CMOS-compatible substrates such as silicon. Ion-slicing has been successfully executed on YIG and other oxide films such as ferroelectric perovskites over the past 20 years (refs. 53,54,71,72). However, the integration of YIG film on a piezoelectric layer has not been demonstrated and this is probably a viable path forward for efficient ME-based IoT devices with high magnetoelectric coupling quality factor (QME).

ME-based memory and logic

Many pathways are being explored to create low-power non-volatile memories within the context of next-generation computing. The push for ultra-low-power logic-in-memory devices builds from the ability of multiferroics to control magnetism with an electric field at room temperature (magnetoelectric control)33. The magnetoelectric switching of a magnetoresistive element can operate at or below 200 mV, with a pathway to get down to 100 mV (ref. 73). A combination of structural manipulation via lanthanum substitution and thickness scaling in multiferroic BFO helped to scale the switching energy density down to ~10 µJ cm−2 and provides a pathway to get to attojoule-class nonvolatile memories (1 aJ per memory bit or logic element), a leap in energy efficiency compared with the pJ per bit requirement for dynamic random-access memory-based technology73. However, the presence of polarization fatigue caused by the degradation of the interface between the CoFe ferromagnetic metal and the oxide BFO harkens back to work on ferroelectric memories, which also suffer from fatigue. To address these degradation issues, it is possible to replace the ferromagnetic CoFe with Sr2(Cr,Re)O6, a double perovskite ferrimagnet, which has proven to be successful if coupled to the BFO layer43. An alternative pathway is to eliminate the ferromagnetic layer and read out the magnon transport in the BFO (and its manipulation with an electric field) using the inverse spin Hall effect (statement based on study currently under review)74.

Logic-memory architectures32 bring together the inverse Rashba–Edelstein effect (IREE)75,76 and electric-field control of magnetism using multiferroic materials. The resulting magnetoelectric spin–orbit-coupled logic device (MESO)34 uses the IREE effect to convert spin to charge (or voltage) and the multiferroic component to perform the opposite conversion of charge to spin24. These devices would be successful if an increase in the IREE voltage output from hundreds of microvoltages (µV) to hundreds of millivoltages (mV) and a reduction in voltage requirement for the magnetoelectric component down to ~100 mV (which is critical for obtaining an aJ per memory bit or logic element) are demonstrated24. Going to switching voltages below 100 mV in ferroelectric and multiferroic states is a challenging research opportunity, which requires the reduction of the coercive field either through pushing the BFO layer to a phase boundary or selecting another multiferroic material. Scaling down to 100 mV also necessitates low switching fields (for example, through careful tuning of the Landau double well barrier height) or that the switching behaviour scales well with reduced thickness. The ultrathin limits (<20 nm) are crucial and require further theoretical and/or experimental exploration.

The speed limit of the switching between ferroelectric and multiferroic seems to be linked to the acoustic phonon mode (that is the velocity of sound in the material, around 3,000 m s−1) tied to lattice deformation during polar state switching, indicating switching times on the order of tens of picoseconds. However, accessing such timescales mandates rapid electronics (for example, sub-ten picosecond pulse generators, high-speed oscilloscopes), explaining limited measurements in this domain77. Lattice dynamics during dipolar switching requires extensive study. Such studies are needed to understand the switching dynamics of both multiferroics and ferroelectrics78.

Another aspect of importance in the ultrathin limit is the stability of the polar state. Such size effects have been extensively studied in classical ferroelectrics79 that are characterized by a suppression of the order parameter as the thickness is scaled down. In BFO, the polar order parameter is suppressed below 4 nm (ref. 78). However, the stability of the individual order parameters, the coupling between them and electric field manipulation of this coupling below 10 nm are still to be studied.

Magnetoelectrics have also been proposed for ME random access memories: ultrathin ferroelectric or multiferroic layers can be used as the tunnel barrier of a magnetic tunnel junction to modify the magnetoresistance of the junction via interfacial spin polarization80,81,82. This approach is limited by the ferroelectric thickness because of the film polarization stability in the ultrathin limit. Alternatively, the magnetic tunnel junction can be directly grown on a bulk ferroelectric substrate83,84,85. Although the magnetoresistance can be tuned via rotating the free ferromagnetic layer, this approach is commercially nonviable owing to the use of bulk ferroelectric substrates.

ME-based magnetometers

Magnetometers are widely used in biomagnetic sensing, navigation, magnetic anomaly detection, communications and so on86. Biomedical magnetometry is an application space that requires the detection of magnetic field signals in frequency ranges of 1–1,000 Hz and amplitudes of ~20 fT to tens of pT to monitor brain, nerve, skeletal muscle, and heart functions87,88,89,90. As of the beginning of 2024, superconducting quantum interference device magnetometers are state-of-the-art in terms of high-sensitivity biomedical magnetometry. A superconducting quantum interference device can detect magnetic signals in the single-digit fT range; however, the cryogenic cooling requirement drastically limits the overall form factor, power efficiency and applications91. Magnetoelectric magnetometers can operate at room temperatures with demonstrated detection of single-digit fT magnetic fields at kHz (C.D. et al., manuscript in preparation) and are, therefore, well suited to lower the cost, power consumption and form factor for biomedical magnetometry systems86. Although ME magnetometers are best-in-class for low-frequency magnetic field sensing, biocompatible implantable devices have yet to be demonstrated; polymer encapsulation may be a path towards in vivo demonstration.

There are three main schemes of magnetoelectric magnetometry: direct detection, frequency modulation and ΔE effect (magnetic field-induced change in Young’s modulus). The corresponding operating principle and noise spectral density (or magnetic noise, in units T Hz−0.5) for each scheme are reported in Table 1 (refs. 18,23,92,93,94,95). Direct detection relies on detecting alternating current (AC) magnetic fields via the direct ME effect — an AC magnetic field-induced alternating magnetostrictive strain translates to a stress variation in the piezoelectric layer. The relative change in electrical polarization can be monitored via an electrical voltage readout. The primary metric that indicates the promise of a direct detection ME magnetometer is the dependence of the ME voltage coefficient on the applied magnetic field. Typically, these ME magnetometers require a direct current (DC) magnetic bias field to push the sensor operation to the peak operating regime of the ME voltage coefficient in response to the applied magnetic field (Table 1). Reduced magnetic noise can be achieved through direct detection on the electromechanical resonance because the ME sensor output voltage is improved for magnetic signals that coincide with the mechanical resonance frequency (with demonstration of magnetic noise down to 400 fT Hz−0.5 (ref. 93) for MEMS ME sensors, and 3 fT Hz−0.5 for bulk ME sensors (C.D. et al., manuscript in preparation)). Additionally, operating the ME sensor in vacuum (for example, keeping the environment pressure below 10−5 mbar) can lead to a further reduction of the noise (owing to less air damping) and, therefore, to an increase in QMR (ref. 96). Using these direct detection methods, on-resonant ME sensors have a limit of detection in the single-digit fT – pT regime. The resonance enhancement, however, comes at the cost of reduced bandwidth. Detecting low-frequency magnetic fields can be challenging because, below resonance and in the low frequency regime, the sensors suffer from 1/f noise93,97. To circumvent operation at mechanical resonance while still maintaining the necessary sensitivity at lower frequencies, active frequency modulation, either magnetic or electrical98, as well as the ΔE effect can be used (Table 1).

Table 1 ME magnetometer noise performance and operating schemes
Full size table

The frequency modulation approach presents two operation schemes: the magnetic and the electric frequency conversion. Magnetic frequency conversion exploits the nonlinearity of magnetostriction versus magnetic field. An advantage of magnetic modulation frequency conversion is that the sum of the modulation and signal frequencies can be selected to be equal to the mechanical resonance of the sensor; thus, resonant operation can be exploited for any arbitrary signal frequency97. The magnetic noise is, therefore, reduced because the resonance enhancement can be exploited99. For example, an ME sensor based on FeCoSiB/AlN heterostructures showed a magnetic noise down to 60 pT Hz−0.5 at 10 Hz (ref. 95). Electric frequency conversion relies on a piezoelectrically induced AC actuation of the sensor, which through the strain-mediated interface, results in an additional magnetic oscillation in the magnetostrictive component with side band generation97. An ME sensor based on an exchange-biased FeCoSiB stack showed a magnetic noise in the low nT Hz−0.5 range94. The electric frequency conversion approach has a smaller form factor and lower power consumption than the magnetic modulation one because it does not require an external driving coil. For both approaches, ME sensors based on frequency modulation are less susceptible to the 1/f noise — which dominates at lower frequencies — and, thus, have improved magnetic noise compared with the passive operation of the sensors100.

The ΔE effect is the change in the Young’s modulus, E, of a magnetic material, when a magnetic field is applied. This change provides additional magnetic field-induced magnetostrictive strain to the elastic Hookean strain101. Upon the application of an external magnetic field, the Young’s modulus of an ME resonator changes because of the ΔE effect and, therefore, a variation in the electromechanical resonance frequency and impedance is observed (Table 1). Self-biased NEMS magnetoelectric sensors based on the ΔE effect have been demonstrated in a FeGaB/AlN nano-plate resonator with a DC magnetic field detection limit down to 300 pT (ref. 19). Additionally, Love wave sensors based on the ΔE effect have a magnetic field dependent phase change caused by the variations in the elastic properties of the magnetic film and, therefore, the wave speed23. Love wave sensors based on FeCoSiB demonstrated a magnetic noise of approximately 100 pT Hz−0.5 over a 50-kHz bandwidth and a dynamic range of 120 dB (ref. 23). A magnetic noise of 28 pT Hz−0.5 at 10 Hz was demonstrated for a ΔE effect Love wave surface acoustic wave sensor based on an exchange-biased stack of FeCoSiB and Ni81Fe19 coupled to an antiferromagnetic Mn80Ir20 layer102. The exchange-biased stack with FeCoSiB coupled to an AFM layer was designed to eliminate domain walls and, therefore, the magnetic 1/f noise was reduced by 8 dB compared with a comparable control device without an exchange bias material stack102. Existing devices enable the detection of magnetic fields at levels of approximately fT Hz−0.5 at 24 kHz (C.D. et al., manuscript in preparation), showing great potential for sensing extremely weak magnetic signals.

ME antennas

Being one of the largest components in electronic systems, antennas have so far been resistant to size miniaturization, leading to electronics with large size, weight, power and cost (SWaP-C). Conventional compact antennas operate at their electromagnetic resonance, leading to antenna sizes comparable to their electromagnetic wavelengths, in the order of centimetres at GHz frequencies. In addition, conventional antennas are based on electric currents that suffer from platform effects, leading to a high profile.

ME antennas are mechanical antennas based on layered ME piezoelectric–magnetostrictive heterostructures (Fig. 2). By applying an AC voltage on the piezoelectric layer, oscillating mechanical strain is induced and transferred to the magnetostrictive film. This electric-mechanical-magnetic energy transduction at the electromechanical resonance triggers AC magnetization oscillation in the magnetic layer, leading to magnetic dipole induced electromagnetic radiation (Fig. 2a). The ME antenna can transmit and receive electromagnetic waves with the same ME heterostructure. Operating at their electromechanical resonance, ultra-compact ME antennas have dimensions comparable to their mechanical or acoustic wavelength, which is five orders of magnitude smaller than the electromagnetic wavelength (Fig. 2b). At the same time, ME antennas are based on magnetic currents, instead of electric currents; thus, ME antennas are immune from the platform effect and exhibit enhanced radiation on a ground plane22,26,31,32,49 (Fig. 2c).

Fig. 2: Smart ME antenna characteristics.
figure 2

a, Magnetoelectric (ME) antenna piezoelectric–magnetostrictive composites. In the transmitting mode, an alternating current (AC) voltage is applied to the piezoelectric layer, generating a strain wave via piezoelectric effect. The strain wave induces the magnetostrictive film deformation and triggers the magnetization dynamics for AC magnetic field radiation (HAC). In the receiving mode, the magnetization in response to the radiofrequency (RF) magnetic fields from electromagnetic waves leads to magnetic film deformation owing to magnetostriction effect. The strain in magnetic film gives rise to the deformation of the piezoelectric layer and generates RF voltage output via inverse piezoelectric effect. The strain-mediated transverse (αE,31) and longitudinal (αE,33) ME coupling leads to in-plane and out-of-plane radiation magnetic field HAC. b, Scanning electron microscopy (SEM) image of ME antenna at very-high-frequency (VHF) and ultra-high-frequency (UHF) bands, showing ultra-low cost, size, weight, and power (CSWaP) feature of ME antennas. c, Illustration of ground plane or platform immunity of ME antennas. The out-of-phase imaging electric current of a conventional electrical antenna mounted on a metallic surface, ground or human body cancels the antenna radiation (left side), whereas ME antenna utilizes magnetic current for radiation and its in-phase imaging current contributes to 3 dB gain enhancement of ME antenna radiation (right side). d, Power handling test of the film bulk acoustic resonator (FBAR) and solid-mounted resonator (SMR) antennas. The FBAR antenna has a 1 dB compression point (P1dB) of 7.1dBm whereas the P1dB of SMR antenna is 30.4 dBm, indicating that SMR antenna shows much better power handling capability. Panel b is reprinted from ref. 22, CC BY 4.0. Panel c is adapted with permission from ref. 27, IEEE. Panel d is reprinted from ref. 31, CC BY 4.0.

Full size image

In 2017, the first ME film bulk acoustic resonator (FBAR) antenna was experimentally demonstrated, marking a significant milestone: it operated at 2.53 GHz and showed an antenna gain of –18 dBi. The structure presented a FeGaB/AlN ME resonating disk with a thickness of ~1 µm and a diameter of 200 µm released but anchored on a 0.5-mm-thick Si substrate22. Mounting a conventional dipole antenna on a ground plane, such as a metallic surface or human body can lead to a detuned antenna impedance and detrimental effects on the performance, as the ground plane can reflect, absorb, and scatter electromagnetic waves and cause destructive interference. ME antennas, however, use magnetic dipole for radiation and show ground plane immunity as the image dipole of the ME antenna on a ground plane is in-phase with the magnetic dipole itself. This ground plane immunity and a ~3 dB gain enhancement on a ground plane have been observed for ME FBAR antennas27. ME antennas are ideally suited as conformal antennas for platform-independent deployments, such as in biomedical health monitoring103.

Despite the superior performance of ultra-high-frequency (UHF) FBAR ME antennas as conformal antennas over various ground plane platforms, they present physical fragility because they are released and suspended devices22 (Fig. 3a). Solid-mounted resonator (SMR) ME antennas utilize acoustic Bragg reflectors to robustly support ME resonators and effectively confine acoustic energy, enabling comparable performance with 24 dB enhanced power handling compared with FBAR ME antennas31,49 (Figs. 2d and 3b). The performance of the ME antenna can be further improved through an array topology: the simulated induced output voltage of the mechanically driven ME antenna can be tripled by using three ME resonators in series compared with a single resonator owing to nonlinear coupling26,104. Additionally, the bandwidth of the ME antenna can be larger through a resonator array with varying resonance frequencies (either through varying the thickness or the width of the resonator, depending on the targeted acoustic mode)104 (Fig. 3c). These UHF ME antennas are self-biased and passive devices with much higher gains than similarly sized conventional antennas, making them ideal for portable and/or wearable IoT devices or biomedical implantable devices26.

Fig. 3: UHF and VLFME antenna architecture.
figure 3

a, Magnetoelectric (ME) ultra-high-frequency (UHF) film bulk acoustic resonator (FBAR) antenna based on FeGaB/AlN resonator. The Si substrate below the resonator was released by XeF2 vapour to form a free vibrating resonator with respect to anchors. b, ME solid-mounted resonator (SMR) antenna based on FeGaB/ZnO resonator on (SiO2/W)3 acoustic Bragg reflectors. c, Very-high-frequency (VHF) and ultra-high-frequency (UHF) dual-band smart ME antenna for magnetic sensing and wireless communication and power transfer. d, ME very-low-frequency (VLF) antenna based on PZT fibres sandwiched by three layers of Metglas on top and bottom. The deformation of PZT fibres is excited by interdigital electrodes (IDE). e, Quartz/Metglas VLF ME antenna array for underground and underwater communication. Panel c is adapted from ref. 26, CC BY 4.0. Panel e is adapted with permission from ref. 41, IEEE.

Full size image

VLF ME antennas (3~30 kHz) have been experimentally demonstrated and patented21,29,30 (Fig. 3d). The VLF signals can penetrate deep into dense and lossy media, such as water, soil and rock. State-of-the-art VLF antennas have dimensions that are miles across and are too large to be implemented in underwater or underground environments. For base station VLF transmitters, these physically gigantic VLF antennas are still electrically small compared with their wavelength, have low radiation resistance and consume megawatts (MW) of power105. In 2020, experimental and theoretical results on ultra-compact (~10 × 1 × 0.1 cm3) ME mechanical antennas and ME communication systems have been demonstrated with distances over 120 m (ref. 30). The maximum transmission distance of a single mechanical antenna is still a few hundred metres within near-field range, restricting long-distance communication30. In 2022, a VLF ME 12-antenna array with a high converse magnetoelectric coupling quality factor (QCME = 120), greater than twice the typical value of 48, was demonstrated, showing an order of magnitude increase in radiation over a single ME antenna29,41 (Fig. 3e). The compact ME antenna array could linearly enhance its radiation B-field strength from 20 nT for a single unit to 210 nT for a 12-element array. Both the radiation magnetic field intensity and the antenna efficiency were linearly boosted using a large number of arrays29.

High-Q VLF ME antennas have a very limited bandwidth, which can be overcome through a nonlinear antenna modulation scheme to potentially achieve a high data rate of 2–10 kbps without the use of a mixer. A data rate that is one order of magnitude higher than that of state-of-the-art VLF communication systems106. Combining the enhanced field strength via array deployment with high data rate via a nonlinear antenna modulation, VLF ME antenna arrays can realize ultra-compact mobile bi-directional VLF communication links between underwater and shore or between underwater and air across the air–water boundary29. VLF underwater and underground sensor networks can, thus, happen with the help of the earth-ionosphere waveguide. Up to 5 m under water, ME antenna arrays in seawater environments show similar transmission behaviours as in air, with no reflection at the seawater–air boundary29. As the equivalent dipole moment amplifies a thousand-fold, over a 30-m transmission distance at 24 kHz becomes achievable29. Compared with other mechanical antennas107, VLF ME antenna performance shows a magnetic noise of ~3 pT Hz−0.5 at 24 kHz and, thus, are extremely sensitive receivers (C.D. et al., manuscript in preparation). Overall, VLF ME antennas have the advantage of ultra-compact size, low power consumption, 10−15 T magnetic field sensing, 10× enhanced data rates, and enhanced radiation strength and efficiency with array deployment compared with alternative mechanical antenna approaches.

ME WPT

WPT enables versatility in applications wherein passive operation is desired, as in the case of IoT sensor nodes installed at locations with limited approachability or for invasive bioimplants. Layered ME heterostructures have been adopted and studied for WPT. In this case, the magnetostrictive material picks up the magnetic field in space and converts it to strain. The strain is then transferred to the piezoelectric layer through ME coupling, giving an electric voltage usable as a power supply108,109,110.

Besides WPT, ME antennas can harvest energy from mechanical vibrations and magnetic fields108,109,110. The magnetic field-based energy harvesting devices directly pick up the magnetic fields in space either in stray form or as the transmitted signal. The WPT efficiency of a ME receiver can reach more than 80% (refs. 111,112), which is comparable to or higher than the resonant coil-to-coil inductive WPT link. At large distances, ME receivers can have efficiencies one order of magnitude higher than the ones obtainable by resonant coil-to-coil WPT links, making ME antennas competitive candidates for large-scale and long-distance WPT networks113.

Taking advantage of their ultra-compact sizes and high efficiency for data transmission and WPT, ME antennas have great potential for bioimplant applications, especially considering the strict size limitations and high energy absorption in tissues at higher frequency114. In this direction, engineering novel materials for device flexibility and biocompatibility115,116, and developing system-level bioimplants for magnetic sensing and stimulation117 have become paramount. For example, NEMS smart ME antennas can perform WPT and magnetic field sensing simultaneously26. The achieved figure of merit for WPT is one to two orders higher than that of miniaturized micro-coils with a magnetic field limit of detection of 300–500 pT between 100 and 1,500 Hz. In another example, a system-level ME bioimplant with an output power of 1.17 mW was obtained under a 6 W transmitter power at an operating distance of 30 mm and was used to produce a 3 Hz stimulation in a rat model103. The subsequent works on ME bioimplants combined data communication and WPT and optimized the modulation schemes for a backscattered communication system118,119. These ME antennas with excellent performance on data communication and WPT enable ultra-compact passive IoT devices for simultaneous wireless information and power transfer operation in a single device, as demonstrated they exploit distinct acoustic resonance modes, enabling simultaneous magnetic field sensing, data communication, and WPT26.

MEs for power, RF and microwave electronics

In the realm of power, RF and microwave electronics, tunable and nonreciprocal components with compact size and superior power efficiency are in high demand. In this section, we focus on magnetoelectric and magnetoacoustic devices and systems that enable tunability and nonreciprocity and exhibit promising compactness, power efficiency and versatility.

ME-tunable components

The demand for tunability in RF devices, driven by size reduction, cost-effectiveness and versatility requirements has led to the exploration of methods such as CMOS switches and varactors, M/NEMS switches and varactors, ferroelectric varactors, and magnetics120. However, these approaches are slow, bulky and power-hungry and often introduce discrete tunable states, structural fragility. ME materials exhibit electric field and magnetic field control of high permeability and permittivity, ferromagnetic resonance (FMR), and non-uniform spin waves, which can be harnessed for tunable power, RF and microwave components, including inductors, filters, phase shifters and resonators120. The strain generated through piezoelectric or electrostrictive effects induces an effective magnetic field, altering magnetic properties and device performance16,17.

Tunable permeability or Faraday rotation121 in ferrites are leveraged to achieve phase shifts through transmission line122,123,124,125,126,127 or waveguide structures128. The Polder tensor, linked to the frequency and magnetic field, dictates the phase shift, thus enabling tunability via piezoelectric–electrostrictive materials. This capability was demonstrated for phase shifters based on a partially biased ferrite YIG, and tunable phase shifters using a YIG/PMN-PT ME laminate124 (Fig. 4a). These examples underscore the potential for ME materials as low-cost, power-efficient, tunable components.

Fig. 4: Magnetoelectric tunable devices for RF, microwave and power electronics and efficient nonreciprocal magnetoacoustic isolator, phase shifter or circulator.
figure 4

a, State-of-the-art tunable phase shifter125 utilizing electric (E)-field tuning permeability in yttrium iron garnet (YIG)/ Pb(Mg1/3Nb2/3)O3–PbTiO3 (PMN-PT) heterostructure. It achieves a phase shift of 119° at 5 GHz and a low insertion loss of 3.5 dB at 5.5 GHz. b, State-of-the-art tunable inductor136 utilizing E-field tuning permeability. Polyimide with FeGaB is used as a magnetic core. The permeability of FeGaB can be tuned by applying voltage on PMN-PT or a bias magnetic field. It achieves an inductance of 1.2~3.5 nH and a maximum inductor quality factor (QL) of 32.7. c, State-of-the-art tunable phase shifter147 utilizing E-field tuning spin wave. The dispersion relation of spin wave in NiZn ferrite tuned by an applied voltage on PMN-PT or a bias magnetic field allows bandpass transmission centred at 3.78 GHz, with an E-field tunability of 55 MHz (kV cm−1)−1 and a magnetic (H)-field tunability of 5 MHz Oe−1. d, State-of-the-art tunable bandpass filter148 utilizing ∆E effect. The applied voltage on the AlN piezoelectric layer or an applied bias magnetic field leads to the Young’s modulus change of the FeGaB magnetostrictive layer. The Young’s modulus change allows for bandpass transmission with an E-field tunability of 2.3 kHz V−1 and a H-field tunability of 5 kHz Oe−1. e, Dispersive relation (left), transmission versus frequency (middle) and transmission versus vector magnetic field (right) of acoustic-driven spin resonance39,166. Acoustic-driven spin resonance shows strong absorption but weak nonreciprocity at the anti-crossing frequency of the spin wave and acoustic wave modes. The corresponding devices incorporate single magnetic layers (Ni, FeGaB, Co, etc.) between interdigital transducers (IDTs) on the piezoelectric substrate, which can be used to efficiently excite quantum defects. f, Dispersive relation (left), transmission versus frequency (middle) and transmission versus vector magnetic field (right) of wideband nonreciprocal transmission25,166. Wideband nonreciprocity leverages the strong coupling between surface acoustic wave (SAW) and non-reciprocal spin wave in synthetic antiferromagnetic (AFM) stack (FeGaB/SiO2/FeGaB, Ru/Co/Ru, etc.), which is used for nonreciprocal isolators, phase shifters or circulators. Panel a is adapted with permission from ref. 125, IEEE. Panel b is adapted with permission from ref. 136, IEEE. Panel c is adapted with permission from ref. 147, IEEE. Panel d is adapted with permission from ref. 148, IEEE. In panel e, dispersive relation and transmission versus frequency graphs are reprinted with permission from ref. 166, APS, whereas transmission versus magnetic field vector graph is adapted with permission from ref. 39, AAAS. In panel f, dispersive relation and transmission versus frequency graphs are reprinted with permission from ref. 166, APS, whereas transmission versus magnetic field vector graph is modified from ref. 25, the authors, some rights reserved; exclusive licensee: AAAS. Distributed under a CC BY-NC 4.0 license.

Full size image

Inductors are ubiquitous circuit elements. To enhance the performance of the inductors and reduce their size, high-permeability materials such as NiFe permalloy129, NiZn–ferrite130, FeCoCu (ref. 131) and FeGaB (refs. 132,133) are used as magnetic cores. Magnetic or insulator laminated cores133,134,135 are used to mitigate eddy currents. However, these inductors are typically not tunable or are tuned via an applied magnetic field, which is not a scalable approach. ME inductors with a magnetic/ferroelectric ME heterostructure core have been demonstrated with voltage-tunable inductance, increased inductor quality factor (QL), and reduced eddy current loss35,136,137,138. Voltage-tunable inductors composed of PZT and Metglas demonstrated a remarkable 450% inductance tunability at 1 kHz (ref. 138), albeit with restricted operating frequencies owing to the large eddy currents in the Metglas. FeGaB/Al2O3/PMN-PT composites yield gigahertz-range inductors with a maximum QL of 32.7% and 191% inductance tunability at 1.5 GHz (ref. 136) (Fig. 4b).

Tunable bandpass or bandstop filters are widely used for frequency selectivity in signal processing. Magnetic material-based bandstop filters leverage the FMR absorption for notch frequency characteristics, using ferrite139,140,141,142,143 or FeGaB (ref. 144). Piezoelectric–electrostrictive materials bonded to magnetic layers enable voltage-based frequency tuning. YIG/PZT heterostructures produced a bandstop filter with 15 MHz tuning at −50 dB in-band rejection145. ME coupling facilitates voltage-tunable bandpass filters via magnetostatic surface wave transmission in ferrites. The magnetic field-dependent dispersion relation of magnetostatic surface waves permits direct or strain-induced magnetic field tuning146. For example, electrically and magnetically tunable ME bandpass filters were demonstrated using NiZn ferrite and PMN-PT slab147 (Fig. 4c). Additionally, ME-based mechanical resonance led to tunable frequency selectivity in free-standing FeGaB/Al2O3 structures148 (Fig. 4d).

Tunable ME resonators are crucial for studying the fundamental electromagnetic coupling in tunable ME devices such as filters, phase shifters, and inductors, which are based on magnetic/ferroelectric or magnetic/piezoelectric multiferroic heterostructures149. Compared with conventional tunable resonators by magnetic field from magnetic cores, these heterostructures with strong converse ME coupling enable effective voltage control of magnetism, providing great opportunities for compact, lightweight, and power-efficient voltage-tunable RF and microwave signal processing devices150. The ME resonators also have competitive tuning range, tuning speed, power handling, linearity, and cost. One notable milestone of the ME resonators is the realization of a novel FeGaB/PZN-PT heterostructures17, which exhibit a record high electrostatically tunable FMR frequency range of 5.82 GHz with a ratio between fmax and fmin of 4.3, and a giant electrostatically induced magnetic anisotropy field of 750 Oe at both DC and microwave frequencies. The giant frequency tunability makes these FeGaB/PZN-PT multiferroic heterostructures attractive for voltage-tunable microwave ME devices.

Nonreciprocal magnetoacoustic RF devices

Integrated nonreciprocal components, such as isolators and circulators, have revolutionized wireless communication and sensing methods151,152,153. Isolators serve as safeguards, preventing back reflections from affecting on-chip high-power amplifiers151,152,153. Meanwhile, integrated circulators, in combination with self-interference cancellation techniques, pave the way for in-band full-duplex operations in upcoming wireless systems and networks151,152,153. However, the integration of full-duplex transceivers based on these integrated CMOS circulators and self-interference cancelers primarily cater to high-power wireless applications and involve considerable DC power consumption ranging from tens to hundreds of milliwatts. The current iterations of CMOS circulators151,152,153 built upon transistor modulation exhibit high power usage (40–200 mW), rendering them less suitable for low-power applications. Additionally, existing integrated CMOS nonreciprocal components encounter bandwidth limitations owing to their dependence on resonant ring structures151,152,153, leading to compromises in power handling when prioritizing wider bandwidths154,155. Innovative integrated nonreciprocal technology featuring superior power efficiency, linearity and isolation is needed to enable low-power applications within the emerging landscape of IoT.

Nonreciprocal magnetoacoustics through the coupling of surface acoustic wave with nonreciprocal spin waves emerges as a novel approach for integrated nonreciprocal RF devices156,157,158,159,160. The synergy between surface acoustic waves and spin waves results in substantial nonreciprocity accompanied by remarkable power efficiency. The hybridization lays the groundwork for on-chip integrated nonreciprocal magnetoacoustic RF components that seamlessly align with CMOS compatibility, ideally suited for both low-power and wideband full-duplex wireless radio systems. The integrated systems hold immense potential for secure communications in the future of IoT, smart cities, smart grids and industrial automation. Notably, these magnetoacoustic devices bring together the functionalities of RF circulators and surface acoustic wave filters, streamlining systems, and radio attributes such as size, cost, complexity and overall efficiency.

The origin of the nonreciprocity of the hybrid magnetoacoustic devices lies in the interplay between surface acoustic waves and spin waves, leading to the emergence of two distinct hybrid magnetoelastic waves through the anti-crossing of magnons and phonons156. Empirical evidence supporting nonreciprocal transmission in surface acoustic wave devices with various single magnetic film layers, such as FeGaB, Ni and Co37, underpins this principle. However, the strength and operational bandwidth of this transmission remains limited (Fig. 4e).

Ongoing research strives to increase both the coupling strength and bandwidth through magnetoelastic interactions between nonreciprocal spin waves and surface acoustic waves157,158,159,160. One such demonstration of nonreciprocity involved coupling surface acoustic waves to a perpendicular magnetic anisotropy stack made of Ta/Co20Fe60B20/MgO/Al2O3 (ref. 161) with interfacial Dzyaloshinskii–Moriya interaction162. The chirality of rotational lattice deformation coupled with magnetization, through magnetic anisotropies, induces pronounced nonreciprocal attenuation in surface acoustic waves, which can enable adaptable acoustic rectifiers, isolators and filters. In contrast to the confined Dzyaloshinskii–Moriya interaction-induced nonreciprocal band (10~100 MHz)163, Ruderman–Kittel–Kasuya–Yosida antiferromagnetic exchange164 offers a wider nonreciprocal transmission band characterized by substantial isolation strength (Fig. 4f). This broader bandwidth response provides a pathway to versatile wideband nonreciprocal RF isolators, circulators, and phase shifters165,166. Experimental validation of this concept entails an antiferromagnetically coupled stack (FeGaB/Al2O3/FeGaB) situated between two interdigitated transducers on a lithium niobate substrate25,28.

Emerging systems

Here we discuss the prospect and promise of emerging systems that should be further explored to improve the performance of magnetoelectric materials and N/MEMS for wireless communication, magnetic sensing and quantum technologies relevant to IoT applications.

Strong magnon–phonon coupling for energy-efficient transduction

Magnons and phonons are integral quasiparticles in the realm of solid-state physics. Magnons, the result of collective electron spin wave excitation, hold immense potential for applications in advanced information processing, communication and sensing. Their attributes, including tunable frequencies through magnetic fields, rapid operation, and extended transduction distances, stem from their high spin density and magnetic dipolar coupling38. Phonons, which denote collections of acoustic vibrations, excel as carriers of coherent and quantum information owing to their long lifetimes within solid-state platforms. The small wavelengths associated with acoustic waves make phonons particularly suitable for on-chip hybrid quantum systems167, with their frequencies conveniently adjustable through geometric manipulation.

Coherent magnon–phonon coupling is a critical mechanism for seamless information transfer between magnon and phonon excitations168. These fundamental quasiparticles have a pivotal role in bridging various other quantum information carriers. The coupling not only holds the promise of revolutionizing information technologies but also sparks interdisciplinary investigations into hybrid multi-physical cavities such as cavity optomechanics, optomagnonics, and magnomechanics169,170. A central aspect of strong magnon–phonon coupling, identified by anti-crossings in magnon and phonon dispersion relations, gives rise to hybrid magnetoacoustic modes termed magnon polarons. The strength of this coupling is quantified by equation (2):

$$C=\frac{{g}^{2}}{{\kappa }_{M}{\kappa }_{P}}=\frac{{\left(\triangle f\right)}^{2}}{4{\kappa }_{M}{\kappa }_{P}}$$
(2)

where g is the coupling strength of magnons and phonons and Δf is the mode-splitting frequency gap. \({\kappa }_{M}\) and \({\kappa }_{P}\) are the loss rates of the magnon and phonon modes, respectively157.

Recent research has focused on robust magnon–phonon coupling with improved acoustic resonance quality factor (Qacoustic) and spin resonance quality factor (Qspin), coupling strength (g) and cooperativity (C). This shift is driven by its potential contributions to magnon spintronics, spin caloritronics, and emerging phenomena such as the spin Seebeck effect, spin pumping, magnon–phonon conversion, and thermal Hall effect171,172. To enhance Qacoustic, Qspin, g and C, researchers are adopting innovative design methodologies. Strategies include reducing energy dissipation, localizing energy in specific coupling regions, and enhancing magnetoelastic coupling. Notable examples include creating Ni nanomagnets on Si substrates to achieve strong coupling using ultrafast pump-probe techniques173 and leveraging perpendicular standing acoustic waves in GGG substrates to enhance coupling with magnons in low-loss YIG films174. In 2020, a galfenol nanograting with a high magnetoelastic coupling constant was introduced, leading to enhanced cooperativity175,176. Techniques such as Gaussian focusing interdigitated transducers in surface acoustic wave devices and Bragg reflectors for magnetoacoustic systems further enhance the coupling177,178,179. A strong magnon–phonon coupling in bulk YIG/AlN ME resonators, demonstrated in 2022, revealed insights into radiation efficiency enhancement with two split magnetoelastic radiation modes180. In the strong coupling region, the radiation efficiency can be enhanced by over two orders of magnitude with a maximum value of around 3.5% (ref. 180). This coupling can also significantly boost radiation efficiency in ME antennas, offering the potential for highly efficient IoT applications.

Key design parameters for robust magnon–phonon coupling are piezoelectric materials with high electromechanical coupling efficiency and magnetostrictive materials with low loss and high magnetostriction. The potential of ion slicing to create high-quality released YIG films for FBAR and SMR antennas, enhancing radiation efficiency, is also evident72,181,182. Additionally, the replacement of AlN with AlScN having six times enhanced piezoelectric constant183 or use of materials with higher magnetostriction constants such as boron or carbon doped FeGa and FeCo alloys56,184,185, spinel NiZnAl–ferrite186 and magnesium aluminum ferrite oxide MgAl0.5Fe1.5O4 (ref. 187) holds promise for further enhanced coupling. These advances can enable highly efficient ultra-compact ME antennas with high radiation efficiency, ME sensors and magnon–phonon quantum transduction devices.

ADFMR for efficient control of quantum defects

Quantum spin defects, such as nitrogen-vacancy centres in diamonds, are captivating owing to their multifaceted potential in quantum technologies: quantum sensing, quantum computing, quantum communication networks188,189,190 and nanoscale magnetic biological sensing191,192,193. The spin states of nitrogen-vacancy centres can be optically detected and manipulated and are sensitive to nearby magnetic fields, enabling them to function as magnon probes194 and potential transducers for converting GHz qubit excitations to optical photons195. Acoustic-driven ferromagnetic resonance through magnetoelectric bulk acoustic wave or surface acoustic wave resonators is an intriguing avenue for nitrogen-vacancy centre manipulation. Its energy efficiency and compatibility with on-chip integration hold promise for quantum communication, quantum sensing, and quantum information science. Electric control over nitrogen-vacancy centres using ME coupling within CoFeB/PMN-PT bulk acoustic wave resonators has been demonstrated40. Applying an external voltage to PMN-PT induced lattice strain in the CoFeB film, altering its magnetic anisotropy, nitrogen-vacancy electron spin resonance and the nitrogen-vacancy centre relaxation rate.

ADFMR has emerged as an efficient approach for localized voltage control of nitrogen-vacancy centres with comparable Rabi frequencies to microwave excitation while consuming far less power196,197. A thin Ni or Co film between interdigital transducers was used for nitrogen-vacancy centre excitation, exhibiting power absorption symmetry peaking at ϕ = 45° (ref. 39). At 1,429 MHz, distinct nitrogen-vacancy photoluminescence emerged, aligned with nitrogen-vacancy centre’s excited state resonance frequency. Cobalt–nitrogen vacancy coupling was notably stronger than nickel, resulting in comparable photoluminescence change with significantly lower electric absorption power. To bolster magnetic biosensing and improve energy efficiency of information transfer for quantum communication, coupling acoustic and spin waves effectively via surface acoustic wave resonators is essential. To reach this objective low-damping, magnetostrictive materials, CMOS-compatible piezoelectric materials, and improved interdigital transducer electrode design are required.

VLF ME antenna arrays for underwater and underground communications

Underwater communications are pivotal for various oceanic activities, encompassing scientific, environmental, commercial, and military ventures. However, achieving high-speed, high-volume underwater wireless communications remains challenging owing to factors such as multi-path propagation, limited bandwidth and rapid signal attenuation198, especially over extended distances. Acoustic waves are predominantly used for underwater communications because they attenuate over a longer decay length in seawater than radio waves. However, acoustic signals suffer from the water-air communication barrier and have other limitations, such as low data rates, communication delays and high energy consumption30. VLF wireless communications between underwater and shore or air have been one-directional only from shore to underwater. The large VLF antenna sizes make mobile underwater VLF transmitters impractical. This prevents submarines or unmanned underwater vehicles from transmitting wireless VLF signals from underwater to shore or air. Overcoming the boundary between water and air for wireless communications has been a persistent challenge199. The ultra-compact mobile VLF ME antenna arrays offer a promising path to future bi-directional water-air or water-shore and underground wireless communications29,30.

Besides wireless communication, ME antennas have excellent performance for wireless power transfer. ME antennas have demonstrated efficiency one order of magnitude higher than resonant coil-to-coil links for wireless power transfer at large distances with significantly improved tolerance of misalignment over inductive coupling and resonant inductive coupling. With the combined advantages, ME wireless power transfer would be an excellent way of powering mobile robots, unmanned underwater vehicles, etc. Moreover, ultra-compact mobile magnetoelectric antennas and microsystems can be used for highly efficient simultaneous wireless information and power transfer, which is critical for IoT systems and allows the transfer of both information and power simultaneously to the end devices.

Outlook

ME devices demonstrate excellent performance in wireless communication, magnetic sensing and energy transduction with good energy efficiency. ME N/MEMS, such as smart ME antennas, exhibit multifunctionality and versatility in diverse IoT scenarios, combining data communication, magnetic field sensing and wireless energy transfer (Table 2). Fully leveraging the potential of ME materials and devices for IoT remains a work in progress.

Table 2 State-of-the-art performance of magnetoelectric components, goals and potential approaches for IoT applications
Full size table

Integrating ME materials with CMOS-compatible processes is a significant challenge for use in IoT. For instance, low-loss magnetic materials, such as yttrium iron garnet, require high processing temperatures exceeding 700 °C, breaching the thermal limits of back-end-of-line CMOS processing. Low-loss magnetostrictive films, such as FeGaB and FeGaC, are promising alternatives with their low deposition temperature and CMOS compatibility.

ME magnetometers exhibit a magnetic sensing capability of fT Hz−0.5 at 24 kHz (C.D. et al., manuscript in preparation). The goal, however, is to extend this sensitivity to a broader frequency range, spanning from 0.01 Hz to 1 kHz, while maintaining detection thresholds on the order of fT. Novel ME sensor designs and new schemes for noise cancellation need to be explored.

Boosting the gain and the radiation efficiency of UHF ME antennas is essential for IoT applications such as WPT, magnetic sensing and communication. Despite experimental demonstration of high-gain UHF ME antennas, the design and simulation of ME antennas for improving their performance is an open challenge. An efficient bulk acoustic wave-mediated multiferroic antenna was designed and simulated using the finite-difference time-domain technique200. However, being that the simulation model was limited to 1D, its abilities to fully capture the spin wave–acoustic wave interactions and, therefore, the ability to design high-performance ME antennas are limited. Two-dimensional and three-dimensional micromagnetic and acoustic simulation platforms need to be developed that capture the complex interactions of spin waves and acoustic waves to optimize the design of UHF ME antenna units and array structures. Additionally, the magnetic damping of the magnetostrictive layer, typically on the order of 10−4 to 10−3, is another concern that degrades the gain and radiation efficiency of ME antennas. Future efforts need to incorporate low magnetic loss (magnetic damping on the order of 10−5) materials, such as yttrium iron garnet, into the ME NEMS antennas, to boost antenna gain and radiation efficiency.

VLF ME antennas ensure robust performance in lossy mediums such as water and ground owing to reduced EM signal attenuation for magnetic field components, and their compact size allows low-frequency operations that reduce energy loss and dissipation. Additionally, VLF magnetoelectric antennas can significantly enhance data transmission rates by over ten times via nonlinear modulation techniques.

As of the beginning of 2024, ME VLF antennas have only demonstrated near-field communication. An open challenge for VLF ME antennas is to demonstrate VLF ME transmitters with radiation in the far field, and subsequent packaging and testing in a seawater environment. Besides, nonlinear coupling among ME antennas in a compact array is also interesting to explore for the potential realization of frequency synchronization and gain enhancement with optimal interspacing and bias conditions. The nonlinear coupling can enable multiple responsive antennas optimally driven by a single active element, which offers promising avenues for potential minimization of ME antenna array system dimensions. Moreover, novel nonlinear antenna modulation techniques with higher-order harmonic sideband signal output would be another interesting direction to further break the limit of data rates and bandwidth of ME antennas. The emerging approaches combining nonlinear antenna modulation and nonlinear coupling among ME antennas hold significant promise for ushering in a new generation of compact, economically efficient ME antenna arrays with wide bandwidth operation and high data rates.

Existing ME antenna technologies have exhibited over tenfold improvements in WPT efficiency over traditional resonant coil-to-coil methods at long distances, but the challenge lies in demonstrating these capabilities in real-world IoT applications over large distances. New ME antennas for WPT need to be designed and optimized, and demonstration of the full strength of ME antennas for IoT has yet to be carried out on WPT, data communication, magnetic and vibration sensing, and so on.

In the sub-field of nonreciprocal magnetoacoustic RF components, the accumulation of nonreciprocal phase surpasses π radians under various field conditions, offering potential for effective circulator development28. The objective is to create highly efficient magnetoacoustic isolators and circulators with low insertion loss, which requires optimal design via theory and simulation, and follow up with experimental verification.

In terms of ME quantum transduction, with efficient control of nitrogen-vacancy centres by ADFMR, future work is oriented towards coherent control of nitrogen-vacancy centres using well-defined spin waves, with an emphasis on low noise metrics. Heterogeneous integration of low-loss magnetic materials in ADFMR devices for manipulation of nitrogen-vacancy centres in a diamond crystal is desired for maintaining spin wave magnitude and propagation distance and enhancing the relaxation time T1 and coherence time T2. The long T1 and T2 ensure that quantum information transferred from acoustic phonons and then stored in magnons can maintain a sufficient lifetime without notable decay or loss owing to relaxation or decoherence, whereas the strong magnitude and long propagation length of spin wave make it possible to realize the long-distance communication of quantum information between various quantum information storage and processing platforms via magnons.