Understanding mantle convection is essential to understand the thermal and chemical evolution of the Earth and to constrain the forces driving plate tectonics. The rheological properties of the mantle are traditionally inverted from surface geophysical data. Radial profiles of the viscosity are thus available but a lot of uncertainties remain.A more detailed model of mantle rheology could be obtained from the knowledge of the constitutive flow laws of mantle phases. A lot of progresses have been achieved to extend the P, T range accessible to rheological studies. However, constitutive flow laws are only available so far for minerals from the upper mantle. More severe is the timescale issue since phenomenological flow laws must be extrapolated over several orders of magnitude to be applied to mantle convection.Recently, a new field has emerged in materials science called multiscale modelling. It allows to link our understanding of a few elementary mechanisms (usually at the microscopic scale) with a behaviour observed at the macroscopic scale. I consider that this offers a ground-breaking opportunity to set a microphysics-based model of the rheology of mantle phases. Much progress has recently been obtained by my group in this direction. A multiscale model of plastic flow consist in modeling:a) the defects responsible for plastic shear at the atomic scale (dislocations);b) their mobility under the influence of stress and temperature;c) their collective behaviour resulting in plastic flow.I propose to build upon those accomplishments and to model the plastic flow of some key phases of the Earth’s mantle: wadsleyite, ringwoodite, MgSiO3 perovskite and post-perovskite to constrain:i) the viscosity contrast between the transition zone and the lower mantle;ii) the viscosity profile of the lower mantle (and understand the origin of the peak of viscosity at mid-mantle);iii) the rheology at the thermal boundary with the core.