Temperature-dependent island size-morphology correlation in Smoluchowski ripening from machine-learning molecular dynamics

Precise fabrication of nanostructures on surfaces requires an atomic-level understanding of surface ripening. Although Smoluchowski ripening (SR) describes the phenomenon where nanostructures migrate, collide, and coalesce, resolving the coupled temperature-dependent evolution of island size and morphology in a many-island ensemble remains challenging. Here, we develop a high-accuracy machine-learning potential (MLP) combined with large-scale molecular dynamics (MD) simulations to examine nanosecond-scale, early-stage nonequilibrium SR-like coarsening of Zn adatoms on the Zn(0001) basal surface over 100–500 K at 12.5% coverage, using three independent 4 ns trajectories at each temperature. We identify a reproducible three-stage evolution: initial clustering and early aggregation, active coalescence, and quasi-steady ripening. Temperature strongly regulates island size and morphology, with higher temperatures yielding larger, less morphologically anisotropic islands. Crucially, we uncover a consistent temperature-dependent island size-morphology correlation and trace it to the coupling between size-dependent thermodynamic preference for higher-coordination compact structures and temperature-enhanced kinetic relaxation through edge diffusion and post-coalescence rearrangement. These results demonstrate the capability of MLP-MD and provide an atomistic framework for understanding the coupled evolution of island size and morphology during early-stage surface coarsening, with significant implications for fields such as catalysis and electronic devices.