1T1M neuromorphic infrared perception based on PbS-QDs decorated IGZO TFT and NbOx Mott Memristor with event-driven property

Biological perception systems are renowned for their ability to process environmental stimuli with remarkable energy efficiency, primarily through spike-based communication and event-driven mechanisms. Inspired by these, in this paper, we propose an infrared artificial neuromorphic perception neuron (IR-ANPN) designed to detect infrared (IR) light, enabling sensory perception beyond the visible spectrum. The IR-ANPN is designed based on a one-transistor-one-memristor (1T1M) architecture, which consists of a PbS quantum dots-decorated Indium Gallium Zinc Oxide (IGZO) thin film transistor (PbS-QDs/IGZO TFT) and an NbO x Mott memristor. The PbS-QDs/IGZO TFT is responsible for detecting infrared light stimuli, while the NbO x Mott memristor converts these stimuli into neuromorphic spikes. The IR-ANPN can operate in an event-driven mode, which means it transmits spikes only when exposed to IR light, ensuring energy efficiency by remaining dormant in the absence of relevant stimuli. The paper also showcases the potential of the IR-ANPN by constructing a neuromorphic infrared detection array. Combined with a spiking neural network (SNN), the system achieves 92 % recognition accuracy on the MNIST dataset by encoding pixel intensity as spiking frequency. The IR-ANPN architecture enhances sensory capabilities through a neuromorphic approach, enabling the detection of non-visible wavelengths with remarkable energy efficiency and laying the foundation for future intelligent sensing technologies.