Neuron-Inspired Biomolecular Memcapacitors Formed Using Droplet Interface Bilayer Networks

Abstract

Brain-inspired (or neuromorphic) computing circumvents costly bottlenecks in conventional Von Neumann architectures by collocating memory and processing. This is accomplished through dynamic material architectures, strengthening or weakening internal conduction pathways similar to synaptic connections within the brain. A new class of neuromorphic materials approximates synaptic interfaces using lipid membranes assembled via the droplet interface bilayer (DIB) technique. These DIB membranes have been studied as novel memristors or memcapacitors owing to the soft, reconfigurable nature of both the lipid membrane geometry and the embedded ion-conducting channels. In this work, a biomolecular approach to neuromorphic materials is expanded from model synapses to a charge-integrating model neuron. In these serial membrane networks, it is possible to create distributions of voltage-sensitive gates capable of trapping ionic charge. This trapped charge creates transmembrane potential differences that drive changes in the system's net capacitance through electrowetting, providing a synaptic weight that changes in response to the history and timing of input signals. This fundamental change from interfacial memory (dimensions of the membrane) to internal memory (charge trapped within the droplets) provides a functional plasticity capable of multiple weights, longer-term retention roughly an order of magnitude greater than memory stored in the membranes alone, and programming-erasure.