Abstract
The formation of metallic dendrites during battery cycling is a persistent challenge for alkali metal-ion batteries, reducing cycle life and posing safety risks. Although surface defects are often implicated in inhomogeneous metal nucleation, the atomic-scale mechanisms by which they promote metal clustering and subsequent dendrite formation remain poorly understood. Here, we use first-principles calculations to investigate how carbon monovacancies (V(C)) influence the clustering behaviour of alkali metals (Li, Na, and K) on graphene - a common basal-plane motif in graphite, hard carbons, and graphene-based anodes. Clusters of Li, Na, and K of varying size (M (n) for n ∈ {1-12}) are characterised on pristine and defective graphene to understand their stability. On pristine graphene, cluster formation is hindered for Li due to the instability of small clusters (n ≤ 3) and significant Li-Li repulsion, and suppressed for K due to weak K-K binding and its larger ionic radius. In contrast, Na exhibits spontaneous clustering, suggesting a higher propensity for dendrite formation even in the absence of defects. The introduction of a V(C) dramatically alters these trends: it stabilises small (n ≤ 3) clusters across all three metals by enhancing binding strength with the surface and modifying charge localisation. For Li, the vacancy overcomes the barrier to early-stage nucleation; for Na, it promotes growth at even lower metal loadings; and for K, clustering becomes locally favoured albeit only for the smallest cluster sizes (n ≤ 3). These results clarify the defect-facilitated pathways to metal clustering, offering atomistic insight that can inform the development of more dendrite-resistant carbon architectures.