Organic electrochemical random-access memory (ECRAM) for flexible and biohybrid neuromorphic systems

Abstract

Analog neuromorphic computers can overcome the limited throughput of von Neumann architectures by parallelizing programming and reading out of artificial neural network (ANN) weights in resistive memory. However, many resistive memory technologies lack the linear programmability required for a parallel “blind update” programming scheme, leading to increased latency during network training. Here, I introduce organic electrochemical resistive memory devices with high linearity, low write noise, and low operating current, enabling accelerated network training via blind updates. Furthermore, the use of an organic active material is highly suitable for flexible electronic systems and integrated biohybrid interfaces. I will show how the organic resistive memory can directly couple with dopaminergic neuron-like cells to constitute a biohybrid synapse with neurotransmitter-mediated synaptic plasticity. We demonstrate both long-term conditioning and synaptic weight recovery, paving the way for combining artificial neuromorphic systems with biological neural networks.