Ferroelectric/electrolyte synergistically modulated IGZO synaptic transistor with tunable plasticity for reservoir computing

Abstract

Electrolyte-gated transistors are promising candidates for neuromorphic devices due to their operational similarity to biological synapses. However, undesirable conductance update is caused by the uncontrollable migration of ions within the electrolyte, thereby limiting their applications in scenarios requiring precise and reliable learning. In this work, an indium-gallium-zinc oxide (IGZO) synaptic transistor was fabricated via a solution process, employing a Li+-doped ZrOx/lead zirconate titanate (PZT) composite as the gate dielectric layer. Ion migration is effectively modulated by the ferroelectric polarization effect of the PZT film, and the IGZO transistor is endowed with tunable synaptic plasticity. Various synaptic behaviors were emulated, including excitatory postsynaptic currents, paired-pulse facilitation, and learning-experience behavior. Based on the volatile and nonvolatile conductance characteristics of IGZO synaptic transistors, a 4-bit encoded reservoir computing system was developed and applied to pattern recognition tasks, effectively reducing the network scale without compromising accuracy. This work demonstrates a ferroelectric/electrolyte synergistically modulated synaptic transistor for the development of efficient and accurate neuromorphic computing.