Flexible Synaptic Transistors with Seafood-Based Biopolymer Composites for High-Accuracy Handwriting Detection

Naturally abundant seafood-based dielectric composites are demonstrated as promising candidates for fabricating sustainable and flexible synaptic transistors aimed at developing flexible neuromorphic systems. The fabricated devices featuring a chitosan and marine gelatin composite dielectric layer, while operating at −10 V, demonstrated good p-channel characteristics while displaying excellent synaptic behavior. The observed synaptic behavior, including the relaxation time constants derived from the pulse paired facilitation, exhibited a remarkable similarity to that of a biological synapse. The fabricated synaptic transistors demonstrated exceptional electromechanical stability during the thorough tests conducted to assess the directional bending performance. Moreover, the developed single-hidden-layer feed-forward artificial neural network (ANN), incorporating weights derived from the transistors, achieved an impressive accuracy of 99.93% when applied to handwritten digits. The applied multidirectional mechanical stress did not influence the performance of the ANN, even after undergoing 100 continuous bending cycles. These impressive results present significant potential for the development of hardware for sustainable and flexible neuromorphic systems.