Bayesian analysis of Ag thin films formation

Abstract

A detailed numerical study of the formation of metallic silver thin films ranged from 8 up to 50 nm on thickness is presented. The topography of these films was imaged by Atomic Force Microscopy, and starting from these images, some surface parameters were obtained. We characterized the root mean square roughness evolution by a simple power-law model with a coefficient α=0.74±0.01 consistent with the theoretical results of Family and Vicsek (1985), Family (1990). Additionally, we considered different models to describe the distributions of the grains’ heights and sizes, and analyzed them via Bayesian statistics and a Markov Chain Monte Carlo numerical method. This Bayesian analysis has been significantly helpful in this work for allowing the study of the models that represent our data best and considering the experimental errors as instrumental data. The results of this analysis suggest an individual grains’ growth followed by a collapse between neighboring grains.