Design of Functionally Stacked Channels of Oxide Thin-Film Transistors to Mimic Precise Ultralow-Light-Irradiated Synaptic Weight Modulation

Abstract

To utilize continuous ultralow intensity signals from oxide synaptic transistors as artificial synapses that mimic human visual perception, we propose strategic oxide channels that optimally utilize their advantageous functions by stacking two oxide semiconductors with different conductivities. The bottom amorphous indium–gallium–zinc oxide (a-IGZO) layer with a relatively low conductivity was designed for an extremely low initial postsynaptic current (PSCi) by achieving full depletion at a low negative gate voltage, and the stacked top amorphous indium–zinc oxide (a-IZO) layer improved the amplitude of the synaptic current and memory retention owing to the enhancement in the persistent photoconductivity characteristics. We demonstrated an excellent photonic synapse thin-film transistor (TFT) with a precise synaptic weight change even in the range of ultralow light intensity by adapting this stacking IGZO/IZO channel. The proposed device exhibited distinct ∆PSC values of 3.1 and 18.1 nA under ultralow ultraviolet light (350 nm, 50 ms) of 1.6 and 8.0 μW/cm2. In addition, while the lowest light input exhibited short-term plasticity characteristics similar to the “volatile-like” behavior of the human brain with a current recovery close to the initial value, the increase in light intensity caused long-term plasticity characteristics, thus achieving synaptic memory transition in the IGZO/IZO TFTs.

Metal-Oxide Heterojunction: From Material Process to Neuromorphic Applications

“…When artificial synapses are stimulated by external light, the number of photocarriers can be increased to tune the device conductance, which is similar to the function of neurotransmitters in the actual nervous system. Therefore, the photoresponse of artificial synapses is weaker in the dark than in sufficient light, which is very consistent with human visual neural behaviors [113]. This helps neuromorphic devices to capture both static and dynamic images, as shown in Figure 5d [114].…”

“…Figure 5b illustrates a method of emulating the human visual nervous system using neuromorphic devices [112]. [113]. (e) Light adaptation.…”