Low‐Power, Electrochemically Tunable Graphene Synapses for Neuromorphic Computing

Mohammad Taghi Sharbati,Yanhao Du,Jorge Torres,Nolan D. Ardolino,Minhee Yun,Feng Xiong
DOI: https://doi.org/10.1002/adma.201802353
IF: 29.4
2018-07-23
Advanced Materials
Abstract:Brain-inspired neuromorphic computing has the potential to revolutionize the current computing paradigm with its massive parallelism and potentially low power consumption. However, the existing approaches of using digital complementary metal-oxide-semiconductor devices (with "0" and "1" states) to emulate gradual/analog behaviors in the neural network are energy intensive and unsustainable; furthermore, emerging memristor devices still face challenges such as nonlinearities and large write noise. Here, an electrochemical graphene synapse, where the electrical conductance of graphene is reversibly modulated by the concentration of Li ions between the layers of graphene is presented. This fundamentally different mechanism allows to achieve a good energy efficiency (<500 fJ per switching event), analog tunability (>250 nonvolatile states), good endurance, and retention performances, and a linear and symmetric resistance response. Essential neuronal functions such as excitatory and inhibitory synapses, long-term potentiation and depression, and spike timing dependent plasticity with good repeatability are demonstrated. The scaling study suggests that this simple, two-dimensional synapse is scalable in terms of switching energy and speed.
materials science, multidisciplinary,chemistry, physical,physics, applied, condensed matter,nanoscience & nanotechnology
What problem does this paper attempt to address?