Work

Antiferroelectric PbHfO3-Based leaky integrate-and-fire neurons for energy-efficient motion perception and localization

Year: 2025

Type: article

Abstract: Artificial neurons are vital components in neuromorphic networks designed to process information with high parallelism and efficiency, mirroring the human brain's capabilities. Phase-change materials are now widely employed in artificial neurons, where excitation-induced resistance variations are leveraged to construct spiking neural networks that emulate neuromorphic computing systems. However, conventional phase-change memory (PCM)-based artificial neurons rely predominantly on temperature (Joule heating) and remain constrained by high power consumption and limited energy efficiency. Here, an antiferroelectric PbHfO 3 material is used to construct a threshold memristor with a Pd/PbHfO 3 /La 0.67 Sr 0.33 MnO 3 structure. The intrinsic accumulated polarization and spontaneous depolarization of PbHfO 3 AFE films simulate neuronal integration and leaky behavior, achieving a low operating current (1 μA), low threshold power (480 nW), and minimal power consumption (10.99 pJ). A leaky integrate-and-fire artificial neuron was then constructed using this device, integrating its spike encoder with a photoresistor to encode external light intensity without requiring additional digital-to-analog or analog-to-digital conversion circuits, thereby effectively mimicking the biological visual system's distance-based response for object localization. Furthermore, by constructing a neuron array, the system is capable of monitoring the motion direction of a light source. This work demonstrates that antiferroelectric-based threshold devices can fulfill the basic functions of spiking neurons and hold great potential in artificial neural networks.

Source: Nano Materials Science

DOI: 10.1016/j.nanoms.2025.11.008