Leaky-integrate-fire and reconfigurable neuron spiking in a field free spin Hall nano-oscillator with a conically magnetized free layer

Abstract

Spintronic oscillators being highly nonlinear have gained immense attention to mimic the neuron spiking behavior in spiking neural networks used for building neuromorphic computing hardware. However, the need for an external magnetic field to realize spintronic oscillators imposes significant limitations on their scalability, tunability, and fabrication complexity. So, in this study, we have realized a bias-field-free spin Hall nano-oscillator (SHNO) using a heavy metal/magnetic tunnel junction (MTJ) heterostructure. The field-free operation is achieved by biasing the free layer of the MTJ into an easy cone regime. This regime arises when the first-order magnetic anisotropy field and the demagnetization field are balanced, and the influence of the second-order magnetic anisotropy becomes predominant. We have explored the oscillation properties of this field-free SHNO, focusing on frequency tunability with current and the output power spectral density through the macrospin and micromagnetic simulations. We have theoretically derived the critical current necessary for the onset of oscillations. Using this field-free SHNO, we demonstrated neuron-like spiking behavior analogous to the Hodgkin–Huxley model by applying a constant DC current. The tunability of spiking frequency in response to input current was also examined. Moreover, we showcased leaky-integrate-and-fire neuron spiking behavior by applying a pulsed current and the reconfigurable nature of neuron spiking under a time-varying ramp current. These spiking behaviors underline the potential applications of this device in the practical realization of brain-inspired neuromorphic computing with the spiking neural network.