Optical components and circuits for signal generation and processing are essential for artificial neural networks (ANNs). We present a four‐layered organic optical waveguide architecture that mimics an ANN. This structure is constructed from flexible organic crystals of (E)‐1‐(((5‐methylpyridin‐2‐yl)imino)methyl)naphthalene‐2‐ol (MPyIN) using an atomic force microscopy cantilever tip‐based micromanipulation technique. By strategically selecting four MPyIN crystal waveguides of varying lengths, bending them into serpentine‐like forms, and integrating them hierarchically, we create interconnected, neuron‐like optical waveguides with six optical synapses. These synapses enable parallel transmission of passive optical signals through evanescent coupling across multiple paths within the waveguides. The feedforward mechanism allows the synapses to split the input optical signal into four diverging signals with different magnitudes. Certain outputs combine passive and active signals due to diverging and converging optical paths. This hierarchical, ANN‐like architecture offers a foundation for developing smart optical neural networks using multiple emissive and phase‐changing organic crystals.

中文翻译：

---

基于模拟神经网络的蛇纹石状有机伪塑料晶体的分层光波导


用于信号生成和处理的光学元件和电路对于人工神经网络 （ANN） 至关重要。我们提出了一种模拟 ANN 的四层有机光波导架构。该结构由 （E）‐1‐（（5-甲基吡啶‐2‐基）亚氨基）甲基）萘‐2‐醇 （MPyIN） 的柔性有机晶体构成，使用基于原子力显微镜悬臂尖端的显微作技术。通过战略性地选择四个不同长度的 MPyIN 晶体波导，将它们弯曲成蛇形形式，并分层集成它们，我们创建了具有六个光学突触的互连、神经元状光波导。这些突触通过波导内多条路径的倏逝耦合实现无源光信号的并行传输。前馈机制允许突触将输入光信号分成四个不同幅度的发散信号。由于发散和会聚光路，某些输出结合了无源和有源信号。这种分层的、类似 ANN 的架构为使用多个发射和相变有机晶体开发智能光学神经网络奠定了基础。