Oxide-based bionic hetero-dendritic neuron with capabilities of Bienenstock–Cooper–Munro learning activities†

Abstract

The close replication of synaptic function is critically important for achieving cognitive computing based on neuromorphic devices. Among the advanced neural learning rules, Bienenstock–Cooper–Munro (BCM) learning rules have been attracting great attention in neuromorphic electronics. Owing to their rich ion dynamic processes, ionic/electronic hybrid devices show great potential in hardware-based neuromorphic systems. Herein, a sodium alginate/graphene oxide hybrid-based electrolyte-gated indium tin oxide (ITO) hetero-dendritic neuron with multi-gate configuration was fabricated. Owing to its unique interfacial protonic hybrid effect, this device exhibited basic synaptic functions. With its proton-related temporal facilitation, this device demonstrated high-pass filter activities, showing potential in image sharpening. It also exhibited an ultra-low power consumption of ∼93.4 aJ for a single synaptic response. Paired-pulse facilitation behavior was mimicked on such a low power consumption. Subsequently, symmetrical Hebbian spike-timing-dependent plasticity and BCM learning rules with regulatory frequency threshold were simulated. Interestingly, a heterosynaptic model was constructed by integrating coplanar gates. Due to the protonic lateral coupling effects, the BCM learning rules could be modulated via the heterosynaptic mechanism. These results indicate the great potential of the present oxide hetero-dendritic neuron in neuromorphic electronics and brain-inspired cognitive platforms.