Stimulus-dependent spiking and bursting behavior in memsensor circuits: experiment and wave digital modeling

Abstract

Biological information processing pathways in neuron assemblies rely on spike activity, encoding information in the time domain, and operating the highly parallel network at an outstanding robustness and efficiency. One particularly important aspect is the distributed, local pre-processing effectively converting stimulus-induced signals to action potentials, temporally encoding analog information. The field of brain-inspired electronics strives to adapt concepts of information processing in neural networks, e.g., stimulus detection and processing being intertwined. As such, stimulus-modulated resistive switching in memristive devices attracts an increasing attention. This work reports on a three-component memsensor circuit, featuring a UV-sensor, a memristive device with diffusive switching characteristics and a capacitor. Upon application of a DC bias, complex, stimulus-dependent spiking and brain-inspired bursting can be observed, as experimentally showcased using combination of a microstructured, tetrapodal ZnO sensor and a Au/SiO_xN_y/Ag cross-point memristive device. The experimental findings are corroborated by a wave digital model, which successfully replicates both types of behavior and outlines the relation of temporal variation of switching thresholds to the occurrence of bursting activity.