Ultraviolet-driven metal oxide semiconductor synapses with improved long-term potentiation†
Abstract
Approaching the limit of Moore's law has presented an urgent challenge for computing hardware innovation, for which a bio-inspired in-memory computing configuration can offer great potential for next-generation hardware of artificial intelligence. A multiterminal synaptic transistor has been widely regarded as an ideal platform to integrate computing capability with memory function. Among the aspects of exploring the artificial equivalence of bio-neural systems, it is essential to realize long-memory retention in this synaptic transistor configuration. Herein, using a low-cost electrospinning technique combined with an efficient nanowire transfer method, we have mainly fabricated bio-inspired synaptic transistors based on one dimensional InZnO nanowires, which can be effectively tuned by ultraviolet (UV) laser to achieve hour-scale long-memory retention of long-term potentiation (LTP). To trigger the nanowire synapses, a 375 nm UV laser with different pulse numbers and light intensities was used, with corresponding measured electrical signals, such as excitatory post-synaptic currents, paired-pulse facilitation and LTPs. Moreover, we achieved improved hand-written recognition accuracy and obtained a sensing-memory system based on experimental measurements. We believe that this study could provide an opportunity to realize bio-resembled artificial systems with long-time memory capability.
Please wait while we load your content...
