Bend‐Resistant and Energy‐Friendly GO‐PVA/PVA Polymer Electret Synaptic Transistors for Neuromorphic Computations
Xiao Fu,Zhihao Liang,Wentao Shuai,Yushan Li,Honglong Ning,Guoping Su,Xubing Lu,Weiguang Xie,Rihui Yao,Junbiao Peng
DOI: https://doi.org/10.1002/adfm.202308127
IF: 19
2023-09-17
Advanced Functional Materials
Abstract:GO‐PVA/PVA polymer electret synaptic transistors with good stability and bending resistance are prepared by exploiting the slow polarization effect of dipoles. The devices can simulate various biological synaptic properties such as short‐term plasticity, long‐term plasticity, and pair‐pulse facilitation. Furthermore, the recognition accuracy of images is simulated based on the constructed artificial neural network, which achieves 86.8% for the MNIST dataset. Polymer electret synaptic transistor is a promising three‐terminal artificial synaptic device. In this work, the electrical characteristics of the composite insulator and transistor are enhanced by modulating the concentration of the 2D nanofiller graphene oxide (GO) and the stacked film structure based on a polyvinyl alcohol (PVA) matrix. The GO‐PVA/PVA polymer electret synaptic transistors before and after dynamic/static bending exhibit typical synaptic characteristics, including short‐term plasticity, long‐term plasticity, pair‐pulse facilitation, spike‐timing‐dependent plasticity, and "learning–forgetting–relearning" features. Importantly, the device exhibits good cycling stability, uniformity and linearity in the potentiation‐depression cycling test, which is beneficial for improving the accuracy of neuromorphic computations. Also, it shows extremely low energy consumption (≈0.32 fJ). The recognition accuracy of images is simulated based on the constructed artificial neural network, which achieves 86.8% for the MNIST dataset. In addition, the devices maintain high recognition accuracy after dynamic/static bending, indicating that the devices are extremely bend‐resistant. The GO‐PVA/PVA polymer electret synaptic transistor is expected to be a potential candidate for neuromorphic computations and electronic skin.
materials science, multidisciplinary,chemistry, physical,physics, applied, condensed matter,nanoscience & nanotechnology