In this paper, a previously established 3D multi-scale simulation chain of plasma deposition process, based on a combination of a direct simulation Monte Carlo (gas phase) algorithm and a kinetic Monte Carlo (kMC) (film growth) code, is improved by the addition of a particle-in-cell Monte Carlo collision algorithm in order to take into account and clarify the role of charged particles. The kinetic Monte Carlo code is also extended with a binary collision approximation algorithm to handle charged particles. This modelling strategy is successfully applied to the growth of TiO2 thin films by means of reactive magnetron sputtering. In order to highlight the effects of negative oxygen ions, two substrate locations are selected: one in the median plane of the targets and another one off the median plane. The model efficiently predicts the densities and fluxes of both charged and neutral particles towards the substrate. Typical results such as particle densities, the discharge current density and ion flux onto the target, and the various substrate locations are calculated. The angular distribution and energy distribution of all involved particles are sampled at these very same substrate locations and the nanoscale modelling (NASCAM) code, implementing the kMC approach, uses these results to explain the morphology of the experimentally deposited coatings. The changes throughout the transition from metallic deposition to stoichiometric TiO2 of the columnar structure of the deposited films is explained by the suppression of the atom diffusion on the growing film due to Ti oxidation. Moreover, the high-energy negative atomic oxygen ions originating from the targets are identified as the origin of the abnormally low inclination of the columnar structure experimentally observed for the oxide mode coatings. Measurements of the normalized energy flux (energy per deposited atom) are experimentally investigated to support and highlight the important role of energetic particles during film growth.