TaOx-based bifunctional memristor for compact leaky integrate-and-fire neuron

Abstract

With the rapid development of artificial intelligence, artificial neurons have received increasing attention as essential components. However, the complexity of existing artificial neuron circuits is adverse to high integration applications due to their complex configuration. Here, we propose a bifunctional memristor, Ag/Ti/TaOx/Pt, which exhibits dynamic and threshold switching performance with different voltage biases. Utilizing these properties, we connected two Ag/Ti/TaOx/Pt devices in a back-to-back configuration to form a leaky integrate-and-fire (LIF) neuron. Moreover, only a one-step lithography process is required to realize a compact neuron with LIF functions. The electrical tests of the artificial neurons were carried out to verify their ability to execute the LIF function and their adaptability to input signals. This work provides a bifunctional device and a highly compact artificial neuron for neuromorphic computing.