One Biristor-Two Transistor (1B2T) Neuron With Reduced Output Voltage and Pulsewidth for Energy-Efficient Neuromorphic Hardware

Abstract

A highly scalable neuron composed of one biristor and two transistors (1B2T neuron) is proposed. The output voltage and pulsewidth in the 1B2T neuron are reduced compared with that in the previously reported one biristor (1B) neuron solely. The approach can greatly enhance the energy efficiency of the neuromorphic hardware by decreasing the energy consumption. To demonstrate the 1B2T neuron, SPICE simulations of 1B2T were performed, reflecting the measured spiking property of the fabricated 1B. The output voltage and the energy consumption were analyzed under various conditions of two transistors (2T) such as threshold voltage and applied voltage. In addition to reducing the energy consumption of the neuromorphic hardware, the output voltage of the 1B2T neuron is adjustable, by controlling the applied voltage to the neuron.