Neuromorphic electronics, which use artificial photosensitive synapses, can emulate biological nervous systems with in-memory sensing and computing abilities. Benefiting from multiple intra/interactions and strong light-matter coupling, two-dimensional heterostructures are promising synaptic materials for photonic synapses. Two primary strategies, including chemical vapor deposition and physical stacking, have been developed for layered heterostructures, but large-scale growth control over wet-chemical synthesis with comprehensive efficiency remains elusive. Here we demonstrate an interfacial coassembly heterobilayer films from perylene and graphene oxide (GO) precursors, which are spontaneously formed at the interface, with uniform bilayer structure of single-crystal perylene and well-stacked GO over centimeters in size. The planar heterostructure device exhibits an ultrahigh specific detectivity of 3.1 × 10¹³ Jones and ultralow energy consumption of 10⁻⁹ W as well as broadband photoperception from 365 to 1550 nm. Moreover, the device shows outstanding photonic synaptic behaviors with a paired-pulse facilitation (PPF) index of 214% in neuroplasticity, the heterosynapse array has the capability of information reinforcement learning and recognition.

使用人工光敏突触的神经形态电子学可以模拟具有内存感应和计算能力的生物神经系统。得益于多重内部/相互作用和强光-物质耦合，二维异质结构是用于光子突触的有前途的突触材料。已经为层状异质结构开发了两种主要策略，包括化学气相沉积和物理堆叠，但是对具有综合效率的湿化学合成的大规模生长控制仍然难以捉摸。在这里，我们展示了由苝和氧化石墨烯 (GO) 前体形成的界面共组装异质双层薄膜，它们在界面处自发形成，具有均匀的单晶苝双层结构和尺寸超过厘米的良好堆叠的 GO。^{13 Jones 和 10 -9} W的超低能耗 以及从 365 到 1550 nm 的宽带感光。此外，该器件显示出出色的光子突触行为，神经可塑性的双脉冲促进（PPF）指数为214%，异突触阵列具有信息强化学习和识别能力。