Capturing atomic wetting dynamics in real time

Atomic-scale wetting plays a crucial role in the fabrication of materials at the nanoscale, yet remains poorly understood under confinement, in which classical capillarity models fail. The growth of metallic nanowires inside multi-wall carbon nanotubes (MWCNTs) presents a particular challenge, as it requires precise control over wetting, nucleation, and vapour-phase condensation. Here, we show that nanowire formation within MWCNTs follows a two-stage mechanism: nucleation at open ends of the multi-wall nanotube, driven by the curved carbon walls, followed by capillary-driven elongation sustained by continuous condensation. Using in situ atomic-resolution transmission electron microscopy (ARTEM) combined with a deep learning convolutional neural network (CNN) capable of differentiating liquid and solid SnO, as well as intermediate SnOₓ phases, we track nanowire growth at the atomic scale. Mechanistically, we reveal that growth requires the formation of a wetting interface (contact angle < 90°) between the liquid precursor and the nanotube wall – a condition that classical models, such as Kelvin and Lucas–Washburn, fail to capture. Instead, these confined systems are dominated by nucleation kinetics, vapour flux, and interfacial condensation, which collectively drive the time-dependent filling observed in our ARTEM experiments. These findings establish a predictive framework for nanowire growth under confinement, providing a scalable approach to advanced nanomaterials where nanoscale wetting drives vapour-phase condensation, with potential implications for the manufacture of catalytic, energy, nanoelectronics, and quantum materials.