Objective-driven morphology control through scanning-probe-inspired postgrowth perturbations

In this study, we combine a kinetic Monte Carlo model of submonolayer growth with scanning-probe-inspired localized postgrowth perturbations to examine morphology control after deposition. This approach offers a route to reshape an already grown surface, but selecting an effective protocol is difficult because improving one morphology feature can degrade another. We define each perturbation protocol by the active scanned fraction, perturbation strength, scan speed, and number of scan repetitions. Bayesian optimization is used as an initial search tool to identify protocols that modify island morphology while limiting undesirable changes in island density, roughness, mean island size, and island-size dispersion. The morphology score rewards controlled changes in island aspect ratio while penalizing increases in island density, roughness, size dispersion, and excessive anisotropy. On the 100 × 100 lattice, the search and independent larger-ensemble confirmation calculations both identify 90% active scanning and a barrier reduction of 0.60 eV as the leading tested condition. Calculations using lateral lattice sizes of 100, 120, and 150 sites nevertheless show that the ranking of strong perturbations is size sensitive, so this condition is not treated as universally optimal. Pass-number calculations on the largest lattice reveal a clear perturbation-dose effect: additional scan passes reduce the overall morphology score and increase island-size dispersion. Collectively, these results show that localized postgrowth perturbations can guide morphology modification in nonequilibrium surface growth, but effective morphology control requires finite and selective intervention rather than prolonged or repeated exposure.