Search for: All records

The use of magnetic tunnel junction (MTJ)-based devices constitutes an important basis of modern spintronics. However, the switching layer of an MTJ is widely believed to be an unmodifiable setup, instead of a user-defined option, posing a restriction to the function of spintronic devices. In this study, we realized a reliable electrical control of the switching layer in perpendicular MTJs with 0.1 nm Ir dusting. Specifically, a voltage pulse with a higher amplitude drives the magnetization switching of the MTJ's bottom electrode, while a lower voltage amplitude switches its top electrode. We discussed the origin of this controllability and excluded the possibility of back-hopping. Given the established studies on enhancing the voltage-controlled magnetic anisotropy effect by adopting Ir, we attribute this switching behavior to the significant diffusion of Ir atoms into the top electrode, which is supported by scanning transmission electron microscopy with atomic resolution. 

Abstract Nb and its compounds are widely used in quantum computing due to their high superconducting transition temperatures and high critical fields. Devices that combine superconducting performance and spintronic non-volatility could deliver unique functionality. Here we report the study of magnetic tunnel junctions with Nb as the heavy metal layers. An interfacial perpendicular magnetic anisotropy energy density of 1.85 mJ/m²was obtained in Nb/CoFeB/MgO heterostructures. The tunneling magnetoresistance was evaluated in junctions with different thickness combinations and different annealing conditions. An optimized magnetoresistance of 120% was obtained at room temperature, with a damping parameter of 0.011 determined by ferromagnetic resonance. In addition, spin-transfer torque switching has also been successfully observed in these junctions with a quasistatic switching current density of 7.3$$\times \;10^{5}$$ $\times {10}^5$A/cm². 

Despite the promise of superior efficiency and scalability, real‐world deployment of emerging nanoelectronic platforms for brain‐inspired computing have been limited thus far, primarily because of inter‐device variations and intrinsic non‐idealities. In this work, mitigation of these issues is demonstrated by performing learning directly on practical devices through a hardware‐in‐loop approach, utilizing stochastic neurons based on heavy metal/ferromagnetic spin–orbit torque heterostructures. The probabilistic switching and device‐to‐device variability of the fabricated devices of various sizes is characterized to showcase the effect of device dimension on the neuronal dynamics and its consequent impact on network‐level performance. The efficacy of the hardware‐in‐loop scheme is illustrated in a deep learning scenario achieving equivalent software performance. This work paves the way for future large‐scale implementations of neuromorphic hardware and realization of truly autonomous edge‐intelligent devices.