Hydrogel-Based Bipolar Synaptic Device for Artificial Neural Networks

We present a bipolar synaptic device based on PAA/PMMA hydrogel. Bidirectional weight modulation, including multilevel positive and negative weights, is realizes within a single device, eliminating the need for differential circuits. This synaptic device, with ultra-low energy consumption of only <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\sim 1.32 \text{pJ}$</tex> per pulse, exhibits linear and symmetric weight adjustments and achieves a 91.3 % recognition accuracy on the MNIST task, highlighting its potential for efficient neuromorphic computing.