CuPc/SiC heterostructure enabling all-optical-controlled artificial synapses for neuromorphic computing and vision.

Optical synapses with the capability of optical sensing and biological synaptic functions are the fundamental components of neuromorphic vision systems and exhibit distinct advantages in image perception, information storage and neuromorphic computing. However, most optical synapses exhibit irreversible optical response characteristics, which significantly limit their application scenarios. This study proposes an all-optically-controlled synaptic device based on a CuPc-SiC heterostructure that exhibits bidirectional optical response characteristics. Optical potentiation and depression behaviors were induced using 680 nm and 350 nm light, respectively. Benefiting from its optical plasticity, four basic logic operations, namely, "AND", "OR", "NOR", and "NAND", are implemented in the device via bidirectional optical inputs. Furthermore, an artificial neural network model was developed, enabling high-precision fingerprint recognition classification. This study provides new insights into the design of bionic visual devices and the development of neuromorphic vision systems.