Control of surface roughness by varying the temperature while a thin film is deposited
Phys. Rev. E 113, 045504 – Published 27 April, 2026
DOI: https://doi.org/10.1103/2hd9-tfbb
Abstract
In thin film deposition from vapor at constant substrate temperature, the surface roughness may reach large values as the thickness increases or as multilayer islands form in the initial growth stages. Here, we use kinetic Monte Carlo (KMC) simulations of four models of thin film deposition with dynamically changing substrate temperature to search for conditions to reduce the roughness of a variety of surface morphologies of applied relevance. At constant temperature, these models lead to: (1) kinetic roughening of the film surfaces; (2) growth of uncorrelated mounds whose widths increase with the temperature; (3) mound coarsening with selection of slopes that increase with the temperature; and (4) growth of multilayer islands on a foreign substrate and their coalescence to form a continuous film. In all simulations, we consider linearly increasing and linearly decreasing temperatures. With models 2 and 4, time-increasing temperature during the deposition produces final surfaces with smaller roughness than those produced by deposition at any temperature in the same range. This occurs because narrow mounds or islands nucleate at the lowest temperatures and, as the temperature increases, the larger diffusion lengths of the adsorbed species lead to rapid filling of the gaps or valleys, allowing faster coalescence of mounds and islands, and consequent surface smoothening. This physical interpretation shows that deposition in time-increasing temperature may be a route for producing smoother films when high densities of islands or mounds are formed at the lowest temperatures (which is the common feature of models 2 and 4). In the simulations of models 1 and 3, as well as with time-decreasing temperature in all models, a similar effect is not observed, but nontrivial evolutions of the roughness (e.g., nonmonotonic variations) may be found, which significantly differ from the growth modes observed in constant temperature deposition. A method to interpolate the roughness evolution of KMC simulations at constant temperatures is also proposed, but for all models, we show that the growth history in time-varying temperature cannot be quantitatively predicted by interpolation of data obtained in constant temperature growth.
