This study explores transient dynamics in the switching process of spin-orbit torque magnetic random-access memory (SOT-MRAM) devices, which are stabilized by in-plane uniaxial magnetocrystalline anisotropy. The authors develop a theoretical model describing interactions between spin torques and effective magnetic fields during magnetization writing.
The framework identifies regions where switching either fails or succeeds. Special focus is given to a “quasi-stochastic” regime, found between deterministic failed and successful switching zones. This regime arises from the interplay of torque-driven and precession-driven magnetization changes during switching.
These modifications exploit transient phenomena to reduce the energy barrier for switching. As a result, SOT-MRAM devices achieve switching with lower currents and faster write speeds compared to conventional designs.
At higher temperatures, the previously unpredictable stochastic regime transforms into a defined "transition band" marked by monotonic and adjustable probabilistic behavior. This enables controlled probabilistic operation.
Through this addition of control mechanisms through electrical inputs, our framework paves the way for the creation of a fast, efficient probabilistic bit (p-bit) for the field of probabilistic computing.
The proposed model and device modifications improve switching efficiency and enable tunable probabilistic operation in SOT-MRAM, advancing the development of fast, energy-efficient probabilistic computing components.