You should measure a bare silicon substrate which is identical as the one you coated graphene. Then measure your graphene sample on the silicon. The Raman signal of the graphene can be extracted by comparing the signals of your graphene sample and the substrate alone.

You question isnt clear.

Do you want to remove noise from you measured data?

Talha Nisar I want to set the baseline for all spectra to zero for intensity comparison and fitting peaks. I do it manually but I wonder is there any specific function for background elimination or not?

Xuhua Wang thank you for your answer. I will appreciate it if you share your idea how to subtruct the base line?

Dear Musaib,

Yes there is a way to reduce noise level in origin. It's explained here

https://www.youtube.com/watch?v=EItYvqOQe5w

But best would be to increase your measurement time to reduce noise. This way detector receives more scattered photon (more counts) and noise will be small.

Usually, you can use the software provided with the Raman equipment. These softwares have many different options including background removal, noise smoothing, peak fitting,...

For the background removal, you usually try to fit a line or polynomial line. The software usually asks you to put the degree of the polynomial function that fits best. In case of graphene on SiO2, it should be rather straightforward (It gets way more complicated if you have the luminescence/background of Cu or Cu oxide for as grown CVD graphene).

If you conducted a map, you should have an option to apply a similar function to every single spectrum of your map.

If you have trouble finding the appropriate function to fit the background, you can also reduce the Raman shift window. Depending on the grating you use, you might have a fairly large Raman shift range while all you want is the 1100-2900 cm-1 range (to have the D, G, 2D bands and enough "space around each peak to have a proper fit/background removal). If the background is still too difficult for fitting a polynomial function, reduce even more the range around 2 peaks and not the 3 main at the time.

I can give you a more specific answer if you use Labspec6 with a Horiba setup.

Generally the background is not that strong for the case of graphene/SiO2 sample. To compare many spectra in one graph you can normalize the curves which exclude this bad-looking data. Otherwise You can use NGS labspec software which gives you very easy way to remove the background.

Good Luck

Dear Muhammad,

Although you can use some softwares to extract the signals as the above researchers suggested, what i normally teach researchets to learn how to manually extract signals from a sample coated on the substrate is to measure the substrate and measure a sample on the substrate, respectively. Then the Raman spectrum intensity from the sample on the substrate is divided by the background signal from the substrate. Then you can obtain the Raman spectrum intensity from a sample alone theoretically. Sometimes you may need to do further work to subtract the baseline. By in mind, the baseline is likely from the photoluminescence which you need to consider too. Hope it is clear.

Best wishes

Xuhua