Self-Powered, Flexible, and Multi-Sensory Artificial Synaptic Devices for Intelligent Object Recognition

The burgeoning fields of artificial intelligence and bio-inspired robotics necessitate the development of advanced artificial synapses that transcend the limitations of conventional unitary and rigid systems. This study presents a self-powered, flexible, multi-sensory artificial synaptic device based on a heterojunction structure of reduced graphene oxide (rGO) and sulfur vacancy-rich ZnIn 2 S 4 (Sv-ZIS), which ingeniously integrates optical perception and tactile sensing capabilities within a single platform. By engineering interfacial voids and defect states, the dynamic photo-carrier trapping and releasing enable the emulation of fundamental synaptic behaviors under optical stimulation, with relaxation time reaching 200 s. Concurrently, the device exhibits a quantifiable mechanical response to strain and external force, originating from the contact potential difference between the Sv-ZIS and rGO layers. Remarkably, all functionalities are achieved at zero bias, ensuring minimal energy consumption. Leveraging the multi-sensory ability, an artificial neural network is constructed for intelligent object recognition based on their optical transmittance and weight characteristics, achieving a high accuracy of 97.67%. This work establishes a novel paradigm for the development of energy-efficient, flexible, and multi-modal neuromorphic devices, paving the way for advanced applications in soft robotics, wearable electronics, and human-machine interfaces.