Spin-transfer torque magnetic memory (STT-MRAM) is gaining attention for its advantages such as non-volatility, fast read/write speeds, and high endurance, making it a promising technology.
This research explores STT-MRAM's potential as a stochastic memristive device that can mimic synaptic functions, by analyzing its behavior under different programming current regimes (low, intermediate, and high).
Simulations indicate that the intermediate current regime is optimal as it minimizes energy consumption while maintaining robustness against variations, suggesting new opportunities for integrating STT-MTJs in efficient cognitive systems.