Energy-Efficient Stochastic Signal Manipulation in Superparamagnetic Tunnel Junctions via Voltage-Controlled Exchange...

Area of Science:

• Materials Science
• Spintronics
• Neuromorphic Computing

Background:

• Superparamagnetic tunnel junctions (sMTJs) are key for stochastic computing in neuromorphic systems.
• Traditional control methods like spin-transfer torque (STT) and spin-orbit torque (SOT) are power-intensive.
• Developing energy-efficient control mechanisms for sMTJs is crucial for advancing computing.

Purpose of the Study:

• To introduce and demonstrate the voltage-controlled exchange coupling (VCEC) mechanism for sMTJ control.
• To evaluate the power efficiency of VCEC compared to conventional methods.
• To explore the potential of VCEC for neuromorphic computing applications.

Main Methods:

• Implementation of the voltage-controlled exchange coupling (VCEC) mechanism in sMTJs.
• Measurement of power consumption for VCEC switching.
• Characterization of the output response of VCEC-controlled sMTJs.
• Integration of VCEC with spin-orbit torque (SOT) for magnetic state manipulation.

Main Results:

• Achieved ultralow power consumption of 40 nW for sMTJ switching using VCEC, 2 orders of magnitude lower than STT-based methods.
• Demonstrated a sigmoid-shaped output response suitable for neuromorphic applications.
• Validated the combined control of VCEC and SOT for enhanced magnetic state manipulation.

Conclusions:

• The voltage-controlled exchange coupling (VCEC) offers a highly energy-efficient method for controlling sMTJs.
• VCEC is a promising solution for low-power probabilistic bits in neuromorphic computing.
• This work presents the first practical demonstration of VCEC in sMTJs, paving the way for future low-power computing architectures.