Circuit-Synergistic SnSe Artificial Synapses With Ambipolar-Like Plasticity for Color Recognition and Dual-Key...

To break the bottleneck of the von Neumann architecture in dealing with high-performance computing tasks, it is crucial to develop reconfigurable and multifunctional artificial synapses to achieve more complex and dynamic functions. However, the unidirectional photoresponse of current optoelectronic devices limits their further advancement. Here, we develop a wavelength-reconfigurable artificial synapse with the ambipolar-like property based on the semiconductor-circuit synergy. By integrating a dual-power circuit with the unique bidirectional photoresponse of tin selenide, an uncommon multidimensional optoelectronic plasticity is achieved through the dual regulation applied by light and voltage. Information from incident light is encoded and mapped to vector changes in the conductance state of the device. This unique functionality enables readily in-memory color recognition through machine learning with a high accuracy of 96.2% for green, blue, and ultraviolet light. Furthermore, we construct a dual-key optical encryption scheme based on light and voltage modulation on an 8×8 array. This work introduces a novel physical principle for synaptic modulation, demonstrating additional possibilities from the convergence of circuit design and optoelectronic devices. It provides a new pathway for developing multifunctional neuromorphic devices and reconfigurable photonic computing hardware.