The polymer binder is a key constituent of silicon-based electrode formulation for lithium-ion batteries. However, its extremely difficult visualization in fresh electrodes becomes completely impossible in cycled electrodes. Here, the combination of scanning transmission electron microscopy-valence electron energy-loss spectroscopy with the use of a specific electrolyte solvent composition (a mixture of dimethyl carbonate and fluoroethylene carbonate, without ethylene carbonate) allows the visualization, for the first time, of the binder in silicon-based electrodes, even cycled up to 100 cycles. Such an observation is possible as the only solid degradation product of this electrolyte is LiF, as confirmed by quantitative Li-7 and F-19 magic angle spinning nuclear magnetic resonance. The electrodes, based on a blend of silicon and graphite, have a very high surface capacity, 6.5 mAh cm(-2), which makes them meaningful for electric vehicle applications. The performance is appreciable, in particular the first cycle efficiency approaching that of commercial graphite electrodes, as well as the capacity retention in full cell, versus LiNi0.5Mn0.3Co0.2O2, 60% after 100 cycles. This work also illustrates the ultimate state of destructuration of silicon after long cycling, which invites new ways of thinking about the design of this material.