Bioinspired Nanofluidic Memristors Based on Polyelectrolyte Conformation for Synaptic Learning and in-Memory Logic Computing.

Abstract

Bioinspired neuromorphic and in-memory computing requires devices that store and process information through ionic dynamics analogous to biological synapses. Here, we report a polyelectrolyte conformational nanofluidic memristor (PCM) that integrates synaptic plasticity, neuromorphic learning, and stateful ionic logic within a single aqueous platform. The device operates through electric-field-driven, reversible conformational transitions of polyelectrolytes confined inside graphene oxide nanochannels, enabling analog conductance tuning, a sharp, tunable switching threshold, and an ON/OFF ratio exceeding 160. These conformational dynamics endow the PCM with rich synaptic functions, including long-term potentiation and depression, multilevel memory retention, and symmetric weight update rules. By directly mapping the experimentally measured potentiation/depression curves into a physical learning model, we demonstrate high-accuracy neuromorphic learning, achieving 97.1% recognition accuracy on the Modified National Institute of Standards and Technology (MNIST) handwritten-digit dataset. Beyond learning, interconnected PCM units perform stateful ionic OR, IMP, and NAND operations, establishing universal in-memory logic within the same ionic platform. This work introduces a bioinspired nanofluidic computing paradigm that unites the adaptive learning of neural networks and deterministic logic, paving the way to scalable and brain-like ionic processors.