▎ 摘　　要

The graphene reinforced nanocomposites have been attracting considerable attention owing to their excellent mechanical and physical properties. Herein we systematically develop a multiscale scheme of graphene-polyisoprene nanocomposites, in which the effective coarse-grained potentials of polyisoprene (PI), graphene, interface derived from the all-atom models using inverse Boltzmann iteration along with pressure correction, strain energy conservation and energy matching method during graphene pull-out testing respectively. Graphene dispersion, PI entanglement network and PI-graphene network structures are quantitively characterized by the interlayer distance distributions of graphene sheets, conformity of PI chains with the entanglement theory and the conformation of polymer chains adjacent to graphene surfaces respectively. Particularly, there are three regimes of reptation dynamics, chain confinement, sufficient network of PI chains in terms of various graphene loading correspond to the dispersion, percolation and aggregation of graphene. In the rubbery region the initial high reinforcement of shear modulus is caused by trapped conformation at high frequency, while at low fre-quencies, the gained conformational freedom leads a relatively low reinforcement, which increases exponentially with a reduced weight fraction. Generally, our multi-scale strategy may be readily extended to other kinds of filling models, and this work could guide the rational design and preparation of nanocomposites.