Demultiplexing through a multimode fiber using chip-scale diffractive neural networks

The explosive growth of global data traffic is pushing conventional single-mode fiber systems toward their fundamental capacity limits. Multimode fibers (MMFs) using space-division multiplexing (SDM) are promising for improved transmission capacity, connection flexibility, and security of data. However, the complex transmission characteristics of MMFs significantly hinder precise mode demultiplexing. Conventional approaches, such as holographic measurements, phase recovery algorithms, photonic lanterns, and multi-plane light conversion, are severely limited by instability, ambiguity, flexibility, or the complexity of the system. In this paper, we demonstrate for the first time a purely optical, chip-scale AI solution for high-mode isolation, speed-of-light demultiplexing of MMF modes using a three-dimensional diffractive neural network (DNN). The DNN is trained with synthetic modal data and fabricated using two-photon nanolithography. It features a compact size of 120μm ×120μm ×80μm and a diffractive structure size of 1μm<2> for the neurons at the hidden layers of the network. The proposed DNN-based demultiplexer achieves a relative modal amplitude error of 6.93% in simulation and 17.96% in experiment. The AI approach of DNN allows for flexible design and overcomes the size and performance limitations of digital-optical demultiplexers. This work paves the way for compact, low-latency optical processors for high-performance demultiplexers and enables scalable, chip-integrated solutions for next-generation fiber optic networks.