Synaptic Microchannels

Abstract

Charged micro-scale fluid channels display intriguing and non-linear transport phenomena. When forced or integrated in an electric circuit, they display spiking and learning behaviour similar to our neurons and synapses. By adding specific surface chemistry to the walls of these channels, a coupling between charge accumulation, surface charge and conductivity opens up a new spectrum of possibilities for neuromorphic engineering. Specifically, intrinsically non-linear charging dynamics of surface reactions with charged reactants are amplified by the current rectifying behaviour of inhomogeneous channel types, such as cones, creating a semi-independent internal state parameter that changes the conductivity. Among other things, these systems can display near non-volatile memristive properties, frequency dependent plasticity and chemically induced Hebbian learning. Because of their high-level resemblance with the underlying physics and chemistry of synaptic transmission, these systems are promising for use in neuromorphic computing, neural implants and brain-computer interfaces.