Silicon is a very good candidate for the next generation of negative electrodes for Li-ion batteries, due to its high rechargeable capacity. An important issue for the implementation of silicon is the control of the chemical reactivity at the electrode/electrolyte interface upon cycling, especially when using nanometric silicon particles. In this work we observed improved performances of Li//Si cells by using the new salt lithium bis(fluorosulfonyl) imide (LiFSI) with respect to LiPF6. The interfacial chemistry upon long-term cycling was investigated by photoelectron spectroscopy (XPS or PES). A nondestructive depth resolved analysis was carried out by using both soft X-rays (100-800 eV) and hard X-rays (2000-7000 eV) from two different synchrotron facilities and in-house XPS (1486.6 eV). We show that LiFSI allows avoiding the fluorination process of the silicon particles surface upon long-term cycling, which is observed with the common salt LiPF6. As a result the composition in surface silicon phases is modified, and the favorable interactions between the binder and the active material surface are preserved. Moreover a reduction mechanism of the salt LiFSI at the surface of the electrode could be evidenced, and the reactivity of the salt toward reduction was investigated using ab initio calculations. The reduction products deposited at the surface of the electrode act as a passivation layer which prevents further reduction of the salt and preserves the electrochemical performances of the battery.