Understanding and designing magnetoelectric heterostructures guided by computation:progresses,remaining questions,and perspectives

Magnetoelectric composites and heterostructures integrate magnetic and dielectric materials to produce new functionalities,e.g.,magnetoelectric responses that are absent in each of the constituent materials but emerge through the coupling between magnetic order in the magnetic material and electric order in the dielectric material.The magnetoelectric coupling in these composites and heterostructures is typically achieved through the exchange of magnetic,electric,or/and elastic energy across the interfaces between the different constituent materials,and the coupling effect is measured by the degree of conversion between magnetic and electric energy in the absence of an electric current.The strength of magnetoelectric coupling can be tailored by choosing suited materials for each constituent and by geometrical and microstructural designs.In this article,we discuss recent progresses on the understanding of magnetoelectric coupling mechanisms and the design of magnetoelectric heterostructures guided by theory and computation.We outline a number of unsolved issues concerning magnetoelectric heterostructures.We compile a relatively comprehensive experimental dataset on the magnetoelecric coupling coefficients in both bulk and thin-film magnetoelectric composites and offer a perspective on the data-driven computational design of magnetoelectric composites at the mesoscale microstructure level.

Graph neural networks for an accurate and interpretable prediction of the properties of polycrystalline materials

Various machine learning models have been used to predict the properties of polycrystalline materials,but none of them directly consider the physical interactions among neighboring grains despite such microscopic interactions critically determining macroscopic material properties.Here,we develop a graph neural network(GNN)model for obtaining an embedding of polycrystalline microstructure which incorporates not only the physical features of individual grains but also their interactions.The embedding is then linked to the target property using a feed-forward neural network.Using the magnetostriction of polycrystalline Tb_(0.3)Dy_(0.7)Fe_(2) alloys as an example,we show that a single GNN model with fixed network architecture and hyperparameters allows for a low prediction error of~10%over a group of remarkably different microstructures as well as quantifying the importance of each feature in each grain of a microstructure to its magnetostriction.Such a microstructure-graph-based GNN model,therefore,enables an accurate and interpretable prediction of the properties of polycrystalline materials.

Strain-mediated voltage-controlled switching of magnetic skyrmions in nanostructures

Magnetic skyrmions are swirling spin structures stabilized typically by the Dyzaloshinskii-Moriya interaction.The existing control of magnetic skyrmions has often relied on the use of an electric current,which may cause overheating in densely packed devices.Here we demonstrate,using phase-field simulations,that an isolated Néel skyrmion in a magnetic nanodisk can be repeatedly created and deleted by voltage-induced strains from a juxtaposed piezoelectric.Such a skyrmion switching is non-volatile,and consumes only~0.5 fJ per switching which is about five orders of magnitude smaller than that by current-induced spin-transfertorques.It is found that the strain-mediated skyrmion creation occurs through an intermediate vortex-like spin structure,and that the skyrmion deletion occurs though a homogenous shrinkage during which the Néel wall is temporarily transformed to a vortexwall.These findings are expected to stimulate experimental research into strain-mediated voltage control of skyrmions,as well as other chiral spin structures for low-power spintronics.

Machine learning in energy storage materials

With its extremely strong capability of data analysis,machine learning has shown versatile potential in the revolution of the materials research paradigm.Here,taking dielectric capacitors and lithium‐ion batteries as two representa-tive examples,we review substantial advances of machine learning in the research and development of energy storage materials.First,a thorough discussion of the machine learning framework in materials science is presented.Then,we summarize the applications of machine learning from three aspects,including discovering and designing novel materials,enriching theoretical simulations,and assisting experimentation and characterization.Finally,a brief outlook is highlighted to spark more insights on the innovative implementation of machine learning in materials science.

MAGNETOELECTRIC RESPONSES IN MULTIFERROIC COMPOSITE THIN FILMS

Multiferroic composite thinfilms of ferroelectrics and magnets have attracted ever-increasing interest in most recent years.In this review,magnetoelectric(ME)responses as well as their underlying ME coupling mechanisms in such multiferroic composite thinfilms are discussed,oriented by their potential applications in novel ME devices.Among them,the direct ME response,i.e.,magnetic-field control of polarization,can be exploited for micro-sensor applications(sensing magneticfield,electric current,light,etc.),mainly determined by a strain-mediated coupling interaction.The converse ME response,i.e.,electric-field modulation of magnetism,offers great opportunities for new potential devices for spintronics and in data storage applications.A series of prototype ME devices based on both direct and converse ME responses have been presented.The review concludes with a remark on the future possibilities and scientific challenges in thisfield.

Excitation and detection of coherent sub-terahertz magnons in ferromagnetic and antiferromagnetic heterostructures

Excitation of coherent high-frequency magnons(quanta of spin waves)is critical to the development of high-speed magnonic devices.Here we computationally demonstrate the excitation of coherent sub-terahertz(THz)magnons in ferromagnetic(FM)and antiferromagnetic(AFM)thin films by a photoinduced picosecond acoustic pulse.Analytical calculations are also performed to reveal the magnon excitation mechanism.Through spin pumping and spin-charge conversion,these magnons can inject sub-THz charge current into an adjacent heavy-metal film which in turn emits electromagnetic(EM)waves.Using a dynamical phase-field model that considers the coupled dynamics of acoustic waves,spin waves,and EM waves,we show that the emitted EM wave retains the spectral information of all the sub-THz magnon modes and has a sufficiently large amplitude for near-field detection.These predictions indicate that the excitation and detection of sub-THz magnons can be realized in rationally designed FM or AFM thin-film heterostructures via ultrafast optical-pump THz-emission-probe spectroscopy.

Author Correction: Graph neural networks for an accurate and interpretable prediction of the properties of polycrystalline materials

The authors became aware that the message passing (via the term D^^(−1/2)A^D^^(−1/2)) between neighboring nodes was not implemented in the layer-wise update function (Equation (1)) due to an error in the original code of the graph neural network (GNN) model.After code modification,the following changes have been made to the original version of this Article.

Field-free spin-orbit torque perpendicular magnetization switching in ultrathin nanostructures

Magnetic-field-free current-controlled switching of perpendicular magnetization via spin-orbit torque(SOT)is necessary for developing a fast,long data retention,and high-density SOT magnetoresistive random access memory(MRAM).Here,we use both micromagnetic simulations and atomistic spin dynamics(ASD)simulations to demonstrate an approach to field-free SOT perpendicular magnetization switching without requiring any changes in the architecture of a standard SOT-MRAM cell.We show that this field-free switching is enabled by a synergistic effect of lateral geometrical confinement,interfacial Dyzaloshinskii–Moriya interaction(DMI),and current-induced SOT.Both micromagnetic and atomistic understanding of the nucleation and growth kinetics of the reversed domain are established.Notably,atomically resolved spin dynamics at the early stage of nucleation is revealed using ASD simulations.A machine learning model is trained based on~1000 groups of benchmarked micromagnetic simulation data.This machine learning model can be used to rapidly and accurately identify the nanomagnet size,interfacial DMI strength,and the magnitude of current density required for the field-free switching.