Close to the wafer surfaces, our platinum diffusion experiments were found to be at gross discrepancy with the predictions of well-established diffusion models. These differences are associated with the ramping-down of the temperature at the end of the diffusion processes. To obtain a consistent model able to explain the experiments reported previously in the literature together with our experiments, energy barriers had to be included for the various reactions rate. For the Frank-Turnbull, kick-out and bulk recombination reactions, barrier heights of 0.55, 0.16, and 0.57 eV were determined, respectively. The newly established model is able to reproduce platinum diffusion for a considerably wider range of experimental conditions than models before