As above, what method can I use? Usually, the large FWHM in a Raman spectrum mingle the close peaks, and cause it difficult to observe some weak peaks close to a strong one (maybe introduced by a second phase or material). Thus, what's the possible method to reduce it? Thank you!
May be if you use focalized microRaman with a little spot you can obtain a better accuracy. Its also depends of the quality of the sample, but for a given sample, the previous method would be useful.
1) Reduce the operating temperature and 2) deconvolute the peaks to find the desired one.
Thank you all for your help!
To Mr. Gallardo: I've tried confocal Raman for my sample, but all in no avail. It seems that confocal method would only increase the spacial resolution (both in xy plane and the z direction), but not the wavenumber resolution. If it does, please give me some further directions on how to achieve this.
To Mr. Sarkar: The operating temperature is 25 C. As for the deconvolution method, I wonder what function is convoluted into the original spectrum? Or, how can I do that?
To Mr. Clarkson: I will try to reduce the slit width in my later experiment, but I just wonder why it might help to increase the wavenumber resolution. Thank you for your uploading files, I will try curve fitting method later.
I agree with John Clarkson and Sujoy Sarkar. Slit plays an important role in that it spatially filters out the stray light in the filter chamber of your spectrometer. A 100-μm slit may give you better spectral resolution (a smaller laser spot will increase the power density, which helps with the signal vs noise ratio, but won't improve the spectral resolution). You can test different slit options by turning on the fluorescent lamps on your ceiling (I attached a Raman spectrum of the fluoresecent lamps in my lab).
A cooled deep-depletion spectroscopic CCD is often used for Raman spectroscopy since it can collect light signal at -60 or -80 deg C, which minimize the thermal noise (the thermal noise at 25 deg C could bury the weak Raman peaks).
There are other factors that may affect the spectral resolution. For example, if you use a fiber for illumination. The fiber itself can generate appreciable background autofluorescence and Raman scattering, which often need to be removed by including a filter after the illumination fiber (a narrower laser peak will give you narrower Raman peaks). In addition, make sure the spectrometer works well. The lenses in it need to be well aligned, and the filter chamber needs to be well sealed to prevent any stray light from leaking out to the grating.
no.. here is what I meant:
signal vs noise ratio is easy to improve, either by using a greater integration time or increasing the laser power or focusing on a small spot (as Dr. P. Grima Gallardo suggested).
what I talked about in the three paragraphs is how to make your wide Raman peaks narrower. A smaller slit may helps. I also mentioned several other things you can try: check if the laser source is clean (if it have a filter), if the measurement temperature is below -60 deg C, if the lenses in the spectrometer are well aligned and if the filter chamber is well sealed.
Hope you can figure it out.
To Wang: A further question on the slit. The slit is usually used to restrict the spot area, so why would it help to reduse FWHM?
(Raman is occupied these days, so I still don't get a chance to try the narrower-slit method, which would seem to be much easier than the other ones you suggested.)
Thank you again!
Bad news updating for Mr. Wang:
The attempt to use a narrower slit did not reduce FWHM at all.
And for Mr. Kuptsov:
I'm thinking these days that the large FWHM (compared with typical IR spectrum) is due to the nature of the sample. To support my assumption, I tested Si wafer on Raman, and it generated a beautiful narrow peak. So the large FWHM has nothing to do with the machine limit. I also asked an engineer specialized at Raman about the same question some days ago, he thought this (undesirable large FWHM) might be related to the nature of my material, since my sample are usually in size of nano scale.
Thus, Mr.Sarkar's suggestion of deconvolution might be the final solution. (You and Mr. Wang both suggested some cooling method, which would not be appliable in my lab.) So my question is what methamatical function should I use to deconvolute the spectrum, or how to get the function by experiment? Is this a possible solution?
Thank you all!
To Mr. Kuptsov:
I can be sure that the widening is not due to fluorescence. The peak position is in accordance with where it should be, which will cancel the doubt of its validity. I didn't not scan the IR spectrum for this sample, because the peaks I concern about lie in a too lower wavenumber zone for our IR.
I'm interested in the deconvolution method you mentioned. Will you please give me more details? Thank you!
Maybe I should have uploaded a spectrum I got. CeO2 has a peak caused by the deficiancy at around 600 cm-1, but totally covered by MnO2 peaks.
If the Si gave your narrow peaks, the wide peaks from your material are probably real. Materials can have a few wide Raman peaks depending on their compositions.
Look at the figure I attached (Fig. 9 of the following paper). The MnO2 spectrum looks like your spectrum. You can get the pure reference spctrum of MnO2 by removing the buffer background (take a buffer spectrum (without MnO2) and subtract the buffer spectrum from your MnO2 spectrum)
http://iopscience.iop.org/article/10.1088/2043-6262/2/2/025014
Resolve the broad Raman peaks by deconvolution technique and the smaller peaks can be resolved. Deconvolution is an inbuilt function in some of the commercially available Raman Spectroscopic systems. Otherwise an be carried out in Matlab.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs