To handle the restrictions of present synthetic neurons in neuromorphic {hardware} implementation, NbOx-based bifunctional memristors are fabricated to assemble oscillatory models and superior neuronal modules. Relying on the voltage biasing circumstances, the NbOx-based memristor can function as both a threshold switching memristor (TSM) or a dynamic memristor (DyM). TSMs are employed to construct oscillatory models and additional reconfigured right into a weighted multi-terminal neuronal module, enabling real-time spatiotemporal summation of enter spikes primarily based on the leaky integrate-and-fire mannequin. This module demonstrated the aptitude to carry out spike summation and multi-weight synergy. Leveraging the gradual resistance change attribute of DyM, a sequential encoder is applied, permitting the system to acknowledge and reply to the temporal order of spiking alerts. Moreover, DyM is built-in into the oscillatory unit to assemble intensification and attenuation neurons, enabling short-term spiking frequency adaptation. The versatile spiking efficiency of our NbOx bifunctional memristor gives a strategic basis for growing synthetic neurons for next-generation bio-inspired spiking neural networks.
