Phase programming in coupled spintronic oscillators (1811.02154v1)

Abstract: Neurons in the brain behave as a network of coupled nonlinear oscillators processing information by rhythmic activity and interaction. Several technological approaches have been proposed that might enable mimicking the complex information processing of neuromorphic computing, some of them relying on nanoscale oscillators. For example, spin torque oscillators are promising building blocks for the realization of artificial high-density, low-power oscillatory networks (ON) for neuromorphic computing. The local external control and synchronization of the phase relation of oscillatory networks are among the key challenges for implementation with nanotechnologies. Here we propose a new method of phase programming in ONs by manipulation of the saturation magnetization, and consequently the resonance frequency of a single oscillator via Joule heating by a simple DC voltage input. We experimentally demonstrate this method in a pair of stray field coupled magnetic vortex oscillators. Since this method only relies on the oscillatory behavior of coupled oscillators, and the temperature dependence of the saturation magnetization, it allows for variable phase programming in a wide range of geometries and applications that can help advance the efforts of high frequency neuromorphic spintronics up to the GHz regime.