Multiferroic Magnon Spin-Torque Based Reconfigurable Logic-In-Memory

Abstract

In-memory computing, utilizing non-volatile memories capable of performing both information storage and logic operations within the same device, holds the promise for empowering artificial intelligence (AI) with significantly reduced energy consumption. Existing logic-in-memory devices that have been implemented operate mainly based on charge transport, a process that inevitably gives rise to heat dissipation. On the other hand, magnons, bosonic quasiparticles carrying angular momentum, can flow through insulators for information transmission with minimal heat dissipation. However, it remains challenging to develop a magnon-based logic-in-memory device mainly due to the lack of efficient approach for electrical manipulation of magnon transport. Here we present a multiferroic magnon spin-torque (MMST) device that uses magnons as information carriers, in which the antiferromagnetic magnon modes can be electrically excited and controlled by the ferroelectric polarization in a multiferroic bismuth ferrite (BiFeO3) thin film with magnetoelectrically coupled antiferromagnetic and ferroelectric orders. In MMST that consists of multiple multiferroic/ferromagnet memory cells positioned along a spin-current channel, information can be written to magnetic bits in parallel by the magnon-mediated spin torque. We show that the ferroelectric polarization can electrically modulate the magnon spin-torque by controlling the non-collinear antiferromagnetic structure. We further demonstrate reconfigurable logic-in-memory operations in single MMST device. Our findings highlight the potential of multiferroics for controlling magnon information transport and offer a pathway towards room-temperature voltage-controlled, low-power, scalable magnonics for in-memory computing.