Lu Han,Dehui Wang,Mengdie Li,Yang Zhong,Kanghong Liao,Yingbo Shi,Wenjing Jie
As the development of neuromorphic computing , it is meaningful to explore the synaptic memristors based on organic materials owing to their advantages in low cost, good mechanical flexibility and strong biocompatibility. In this work, organic pentacene thin films have been deposited on different substrates by the thermal evaporation method. The prepared thin films on SiO 2 /Si substrates demonstrate the co-existence of digital and analog resistive switching (RS) behaviors as well as a series of synaptic plasticity. To characterize the flexibility of the organic thin films, lateral memristors have been fabricated on polyethylene terephthalate (PET) substrates and demonstrate comparable RS behaviors when the bending radius is decreased to 1.26 cm or after 1000 bending. Furthermore, vertical memristors with Pt/pentacene/indium tin oxide (ITO) structure have been constructed when the organic thin films are deposited on ITO/PET substrates, exhibiting bipolar RS behaviors with low SET voltage of 0.65 V. The vertical synaptic memristors can be used to simulate the “learning-forgetting” experience of human brain. More importantly, an artificial neural network (ANN) can be simulated based on the vertical synaptic memristors, demonstrating 93.8 % recognition accuracy for hand-written digit image. This study suggests potential applications of the organic pentacene thin films for neuromorphic computing.
A pivotal characteristic of neuromorphic systems lies in their inherent capacity for self-adaptation, facilitating environmental adaptation through the attenuation of response to repetitive stimuli. Existing electrolyte-gated synaptic transistors face challenges in simulating these self-adaptive behaviors of the brain. In this work, flexible indium zinc tin oxide (IZTO) nanowires-based synaptic transistors exhibiting self-adaptive behavior were fabricated for the first time via a facile electrospinning technique. The prepared transistors can realize typical synaptic functions by continuously modulating the conductance, including excitatory/inhibitory postsynaptic current and short/long term potentiation. These properties are well maintained even under a bending radius of 12 mm. Additionally, the neuromorphic system based on IZTO nanowires-based transistors achieves a high recognition accuracy exceeding 90 % for hand-written digits recognition. The synaptic transistor exhibits obvious self-adaptive behavior and a low energy consumption with reduction of 8.2 %. Self-adaptive neuromorphic flexible transistors can not only enhance their resemblance to biological systems, but also advance the development of wearable devices and neuromorphic calculation.
危险品化学品经营许可证（带存储）