Na-doped Ga2O3 electrolyte-gated synaptic transistors for neuromorphic computing

Abstract

Ultra-wide bandgap Ga₂O₃ has emerged as a promising material in the emerging field of neuromorphic computing due to its unique optoelectronic properties. However, previous reports on Ga₂O₃-based neuromorphic devices focused mainly on the two-terminal structure, which restricts its broader application. In this work, Na-doped Ga₂O₃ three-terminal electrolyte-gated synaptic transistors (EGSTs) are proposed to realize neuromorphic computing. The Na-doped Ga₂O₃ electrolyte is fabricated via a low-temperature aqueous solution route. Typical synaptic behaviors were successfully emulated, including excitatory/inhibitory post-synaptic currents (EPSC/IPSC), paired-pulse facilitation/depression (PPF/PPD), short-term memory (STM), long-term memory (LTM), high-pass filtering, and spike-number-dependent plasticity (SNDP). The synaptic functions of the Na-doped Ga₂O₃ EGSTs originate from the migration of Na ions within the electrolyte under gate bias. Furthermore, the pattern recognition capability of the Na-doped Ga₂O₃ EGSTs is demonstrated in an artificial neural network (ANN) with a high recognition accuracy. This study highlights the significant potential of Ga₂O₃-based three-terminal synaptic transistors for neuromorphic applications.