Solution-processed LiGdO solid electrolyte for an In2O3 synaptic transistor and its application in neuromorphic computing

Abstract

Emulating biological synaptic behavior using an electrolyte-gated synaptic transistor (EGST) is promising for neuromorphic applications. EGSTs based on solid electrolytes exhibit excellent stability and compatibility with semiconductor manufacturing processes. In this work, lithium doped Gd₂O₃ (LiGdO) solid electrolyte was fabricated by a solution process, and the In₂O₃ EGST based on LiGdO solid electrolyte was integrated. The synaptic functions of the In₂O₃/LiGdO EGST were realized by the migration of lithium ions within the LiGdO electrolyte under gate bias. By modulating the presynaptic spike scheme, the basic functions of biological synapses, including the excitatory postsynaptic current (EPSC), paired-pulse facilitation (PPF), short-term memory (STM), long-term memory (LTM), and high-pass filtering characteristics, were simulated. Based on the potentiation/depression characteristics of the channel conductance, an artificial neural network (ANN) was used for pattern recognition, and a high recognition accuracy of 93% was obtained for the In₂O₃/LiGdO EGST by simulation training. This study offers an opportunity to realize CMOS integration for EGSTs.