Design of concrete structures involve the consideration of their exposure to predict their service life from engineering or more sophisticated models if the effective diffusion coefficient is known. This parameter is generally obtained from natural diffusion or migration tests on concrete samples. Because of calibrations, most of the existing models models do not link the effective diffusion coefficient to the physical and chemical properties of the material at the molecular scale which affect the diffusivity in the gel pores of hydrates. In this project, we propose a multiscale model to predict the diffusion coefficient in cement pastes with partial substitution of blast furnace slag and metakaolin, for which the main hydrated products are calcium aluminosilicates hydrates (C-A-S-H). At first, we will study the diffusion of ions in gel pores on molecular models of C-(A)-S-H considering different pore solutions (bi-species and multispecies), different pore sizes, and different C/S and A/S ratios to cover the ranges observed in SCM blended cementitious materials. Scanning electron microscope image analysis coupled with energy-dispersive X-ray spectroscopy, together with 29Si and 27Al nuclear magnetic resonance (NMR) measurements will validate the molecular models. The molecular approach will highlight the electrical double layer (EDL) phenomenon, whose effect on diffusion has been demonstrated. The influence of the EDL will be quantified experimentally using different methods of zeta potential measurement. Secondly, the nanoscale (~100 nm) will be taken into account with a nanoparticle interaction model. Several representative elementary volumes will be created to reproduce experimental observations from mercury intrusion porosimetry, nitrogen adsorption and proton NMR. Finally, based on a hydration model, we will make a transition to the scale of a representative elementary volume of cement paste using homogenization methods. Migration tests will permit to compare the diffusion coefficients calculated by our multi-scale model. The ambition of this project is to improve the fundamental understanding on the diffusion of ionic species in the cement paste and to answer the following question: how does the composition of the binder affect the diffusion in concrete at different scales? This study could pave the way for better concrete design with a view to a more accurate prediction of the service life of structures.