Synaptic transistor implementing potentiation or depression via field distribution modulation of dual dielectric layers by single-polarity pulsed voltage stimuli

Abstract

Understanding and simulating the synaptic plasticity mechanisms of the human brain are crucial for advancing neuromorphic computers for artificial intelligence (AI) and humanoid applications. This study designs, fabricates, and characterizes a dual-gate dielectric synaptic device that utilizes interfacial charge traps to mimic long-term potentiation (LTP) and long-term depression (LTD), which are essential for learning and memory. The device is fabricated using a standard complementary metal-oxide-semiconductor process and is composed of a TiO2 channel deposited over a dual dielectric of SiOx on Al2O3. Biological LTP and LTD occur in the hippocampus and involve complex neurotransmitter interactions, notably through N-methyl-D-aspartate and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors. In this device, the thickness of the Al2O3 layer determines the synaptic behavior; increasing the Al2O3 thickness increases the voltage across Al2O3 but decreases the voltage across the SiOx layer. Therefore, the electric field distribution determines whether conduction is primarily driven by trapped holes from the charge-storage layer or electrons in the TiO2 channel. Also, by simply varying the applied voltage pulse widths, a single device comprising a thin 10 nm Al2O3 layer can display potentiation or depression with voltage pulses of the same polarity. These results indicate that our device has the potential to function as a synaptic device, effectively replicating brain-like processes and thereby advancing cognitive computing and AI applications.

Keywords

• Synaptic devices
• Dual-gate dielectric
• Charge trap-detrap
• Long-term potentiation
• And long-term depression