Assessment of trapping layer control in IGZO/Al2O3/Ga2O3 synaptic transistor for neuromorphic computing

Abstract

We conducted research to create reverse synapse plasticity using metal oxide semiconductor-based field-effect transistors. Specifically, we used IGZO as the channel, Al2O3 as the tunneling layer, and Ga2O3 as the trapping layer. We adjusted the thickness of the Ga2O3 trapping layers and examined changes in roughness and density. Through this examination, we confirmed the variations and mechanisms of synaptic behaviors with respect to the properties of Ga2O3. We found that controlling the charge traps as functions of pulse time, input voltage, and initialization is key to approaching optimal device conditions. As a result of our research, we obtained a maximum nonlinearity factor of ν = 0.27 for G of synaptic plasticity. This high degree of linearity, particularly near zero nonlinearity, is significant for neuromorphic research in pattern recognition.