You can try to eliminate the 520 cm-1 peak using cross polarization both of on entrance and exit of the scattered beam. But you won't be able to complete eliminate the peak, the cross polarization will atenuate it.

Since I do not manipulate the Raman device by myself, all I know is that, there is , the laser power, the wavelength of the laser, and a collection time of the outgoing beam which can influence the thin film penetration laser. My question: How to work on all these parameters to avoid reaching the substrate according to the thickness of the SiC layer deposited onto monocrystalline silicon substrate?. Thank you.

You have to change the wavelength of the laser you are using. Try a UV laser with less penetration depth in Si.

Well, as Vasil said, you can use a UV laser, but that's not enough if you're measuring a  thin film. The thickness of your measured film should be large enough so that the laser does not reach the substrate. If you have a film in the nm scale, you won't be able to completely eliminate that peak. You could try other Raman techniques, such as SERS or the elusive TERS, but those have their own complications.

Dear Jose Luis Enriquez, first thank you for your help. Do you have references or documentation about this (Uv laser Vs Thin films, SERS and TERS and complications ;) ) Thank you!!!!

Perhaps a stupid question, but can you measure the signal from a reference substrate and then fit this to your data as a baseline by normalizing the signal to the peak wavelength value (of the baseline) to your measured thin film sample.  Wouldn't this give you a relevant peak substraction?  Alternatively, can you choose an excitation wavelength range which does not excite the substrate but still excites your sample?