I'm not 100% sure, but with my own CNT samples it helped to turn the raman laser to an excitation energy of the CNT (resonance condition), which may work for the graphene as well.  Even without a continuous source choosing a wavelength close to the resonant condition helps a lot.  I don't see a reported wavelength so maybe it can help (it would maximize the signal noise ratio)

I've seen papers in the past, but have no experience with SERS:  they deposited a specific colloidal gold onto the graphene and then had enhanced imaging (the SERS range is on the order of nanometers so with even a few layers of graphene the substrate can't be seen relative to the SERS signal of graphene

We use a 532nm Laser. The perfect wave length for graphene on copper would be 514nm, so we are very close to that. The idea with the gold is quite nice. But again, this is would be very time intensive. Best way would be to just optimize the Raman part on as-grown copper. But I guess the wave length is the most important parameter

Dear Jan are you sure the observed background is due to copper? It seems not really due to elastic scattering but rather some kind of luminescence. I would try to measure a naked substrate to confirm. If the background is really intrinsic to copper substrate I see no way of reducing it, you can just do baseline subraction. 

Dear Jan,

Its highly impossible to avoid the Cu background during the acquisition of Raman characterization, due to fluorescence from the annealed Cu foil, But most of the cases excitation wavelength also matters, most suitable wavelength is 514 or 532 nm, to avoid Cu SPR peak is around ~ 450 nm...., one may measure with 633 nm also..

@Mauro: I am certain that it is the copper background. If I measure the transfered samples with the same set up I have clear peaks without any of the high background intensity. Baseline subtraction is very risky because of the weak 2D-Peaks of some samples. By doing so, it is possible that I delete some smaller peaks.

@Ravi: What is the advantage of the 633nm Laser? I thought the noise ratio would increase if I leave the window of 514-532nm.

Extendung exposure time may be helpful. But, if the 2D peak is nearly invisible, can you conform this is graphene?

That's the point why I want to get rid of the intense background. It is possible that the 2D-Peak is there but just very weak and I don't see it. Most of the times I see a broadened peak at around 2700, so I know that there is something there. Increasing the expsoure time has some limits too. I am seeing good results at 60s

Hi Jan,

Did you try doing confocal Raman imaging? This would increase the SNR of both signals, making the subtraction from background not that noisy.

Unfortunately we don't have the possibility to do confocal Raman imaging, but good advice. Thank you!

What about increasing your exposure time and using a grating that centers your reading with a higher resolution around the 2D peak?

Yeah that is exactly what we will do in future measurements. First a quick run with a bad SNR to determine if there are any peaks of interest and then eventually change the grating and increase the exposure time. I guess we just reached the limits for raman on copper and there is not much more possibility of improvement. Thank you very much for your ideas

Hello Manav,

thank you for your interest in my question. I know that the copper background will be diminished with smaller wavelengths. There is an interesting publication about that:

https://www.researchgate.net/publication/256856135_Resonant_Raman_spectroscopy_of_graphene_grown_on_copper_substrates

Even though the background of the copper foil will be smaller, the peaks will be too. Thus, I could lose information about peaks. So I guess the best way for Raman on copper is to use wavelengths around 514-532 nm to get the highest peak intensities. This comes with the higher SNR though, but at least you can see small peaks if you choose a longer integration time.

Article Resonant Raman spectroscopy of graphene grown on copper substrates

Hii 

Its better to use 532 (green laser ) than you can eaisly see the raman peak and dont forget to do Baseline here with i attached with and without background correction Graphene raman on copper using 532nm Laser