In-plane magnetization switching characteristics of energy-efficient strain-mediated nanomagnets assisted by the spin Hall effect at room temperature

Abstract

Magnetization switching of nanomagnets induced by voltage shows an energy-efficient way to spintronic devices. In-plane magnetization switching can be realized by simulating the magnetization dynamics under a combination of spin-orbit torque (SOT) and strain. The magnetization dynamics were simulated using the Landau-Lifshitz-Gilbert equation under macrospin approximation in the presence of thermal noise. Results show that a relatively small voltage (~60 mV) along with a charge current (1011 A/m2) can drive the magnetization flipping 180° (the switching probability near ~0.98), the energy required is ~0.25 pJ. The energy dissipation of Strain + SOT clocking scheme is 4 orders of magnitude less than that of SOT clocking scheme (~4.44 nJ). Low current density (<109 A/m2) makes no contributions to magnetization switching, which was dominated by the thermal noise (the probability ~0.5). If the current is too large, the probability of 180° flipping will not increase significantly when increasing the current, but the power consumption of the device is greatly increased. We should take a compromise between performance and energy dissipation. Moreover, the stochastic switching behavior of the SOT strain-mediated nanomagnets demonstrates the feasibility to mimic the artificial neurons that can be used to construct an artificial neural network to recognize the handwritten digits. The accuracy of the Strain + SOT neurons-based ANN is ~98%, which can work as well as traditional neurons. These results prove that Strain + SOT clocking scheme provides an additional feasible route to realizing energy-efficient spintronic neurons and memory.