Hybrid coupling systems consisting of transition metal dichalcogenides (TMD) and plasmonic nanostructures have emerged as a promising platform to explore exciton–plasmon polaritons. However, the requisite cavity/resonator for strong coupling introduces extra complexities and challenges for waveguiding applications. Alternatively, plasmonic nano-waveguides can also be utilized to provide a non-resonant approach for strong coupling, while their utility is limited by the plasmonic confinement-loss and confinement-momentum trade-offs. Here, based on a cavity-free approach, we overcome these constraints by theoretically strong coupling of a monolayer TMD to a single metal nanowire, generating ultra-confined propagating exciton–plasmon polaritons (PEPPs) that beat the plasmonic trade-offs. By leveraging strong-coupling-induced reformations in energy distribution and combining favorable properties of surface plasmon polaritons (SPPs) and excitons, the generated PEPPs feature ultra-deep subwavelength confinement (down to 1-nm level with mode areas ~ 10–4 of λ2), long propagation length (up to ~ 60 µm), tunable dispersion with versatile mode characters (SPP- and exciton-like mode characters), and small momentum mismatch to free-space photons. With the capability to overcome the trade-offs of SPPs and the compatibility for waveguiding applications, our theoretical results suggest an attractive guided-wave platform to manipulate exciton–plasmon interactions at the ultra-deep subwavelength scale, opening new horizons for waveguiding nano-polaritonic components and devices.

Retrieving waveguiding properties of plasmonic metal nanowires (MNWs) through numerical simulations is time- and computational-resource-consuming, especially for those with abrupt geometric features and broken symmetries. Deep learning provides an alternative approach but is challenging to use due to inadequate generalization performance and the requirement of large sets of training data. Here, we overcome these constraints by proposing a transfer learning approach for modeling MNWs under the guidance of physics. We show that the basic knowledge of plasmon modes can first be learned from free-standing circular MNWs with computationally inexpensive data, and then reused to significantly improve performance in predicting waveguiding properties of MNWs with various complex configurations, enabling much smaller errors (~23–61% reduction), less trainable parameters (~42% reduction), and smaller sets of training data (~50–80% reduction) than direct learning. Compared to numerical simulations, our model reduces the computational time by five orders of magnitude. Compared to other non-deep learning methods, such as the circular-area-equivalence approach and the diagonal-circle approximation, our approach enables not only much higher accuracies, but also more comprehensive characterizations, offering an effective and efficient framework to investigate MNWs that may greatly facilitate the design of polaritonic components and devices.