Magnetoelectric Memory Based on Ferromagnetic/Ferroelectric Multiferroic Heterostructure

Modeling the Converse Magnetoelectric Effect in the Low-Frequency Range

This article is devoted to the theory of the converse magnetoelectric (CME) effect for the longitudinal, bending, longitudinal-shear, and torsional resonance modes and its quasi-static regime. In contrast to the direct ME effect (DME), these issues have not been studied in sufficient detail in the literature. However, in a number of cases, in particular in the study of low-frequency ME antennas, the results obtained are of interest. Detailed calculations with examples were carried out for the longitudinal mode on the symmetric and asymmetric structures based on Metglas/PZT (LN); the bending mode was considered for the asymmetric free structure and structure with rigidly fixed left-end Metglas/PZT (LN); the longitudinal-shear and torsional modes were investigated for the symmetric and asymmetric free structures based on Metglas/GaAs. For the identification of the torsion mode, it was suggested to perform an experiment on the ME structure based on Metglas/bimorphic LN. All calculation results are presented in the form of graphs for the CME coefficients.

Bayesian neural networks using magnetic tunnel junction-based probabilistic in-memory computing

Bayesian neural networks (BNNs) combine the generalizability of deep neural networks (DNNs) with a rigorous quantification of predictive uncertainty, which mitigates overfitting and makes them valuable for high-reliability or safety-critical applications. However, the probabilistic nature of BNNs makes them more computationally intensive on digital hardware and so far, less directly amenable to acceleration by analog in-memory computing as compared to DNNs. This work exploits a novel spintronic bit cell that efficiently and compactly implements Gaussian-distributed BNN values. Specifically, the bit cell combines a tunable stochastic magnetic tunnel junction (MTJ) encoding the trained standard deviation and a multi-bit domain-wall MTJ device independently encoding the trained mean. The two devices can be integrated within the same array, enabling highly efficient, fully analog, probabilistic matrix-vector multiplications. We use micromagnetics simulations as the basis of a system-level model of the spintronic BNN accelerator, demonstrating that our design yields accurate, well-calibrated uncertainty estimates for both classification and regression problems and matches software BNN performance. This result paves the way to spintronic in-memory computing systems implementing trusted neural networks at a modest energy budget.

Analysis of Dynamic Magnetoelastic Coupling in Mechanically Driven Magnetoelectric Antennas

Mechanically driven magnetoelectric antennas are a promising new technology that enable a reduction in antenna size by many orders of magnitude, as compared to conventional antennas. The magnetoelastic coupling in these antennas, a phenomenon playing a direct role in determining performance, has been modeled using approaches that are severely lacking in both accuracy and tractability. In response to this problem, we take a physics-based approach to the analysis of magnetoelastic coupling. We find that certain directions of applied stress will maximize the coupling and we derive general expressions to quantify it. Our results are applied in comprehensive simulations that demonstrate the dynamic nature of the coupling as well as the impact of various operating conditions and material properties. Our work contributes analytical expressions and associated insight that can serve not only as guidelines for the design of mechanically driven magnetoelectric antennas, but also as stepping stones towards the development of more accurate models.