Such puzzles happen from time to time in FTIR. And if you concentrate on such problem - you will find a number of explanations. my professional advice is - do not pay attention - you will lose your time. Do 2--3 more measurements and if the behavior of the baseline repeats - just neglect it

Interested, following.

AS just mentioned above Do it 3 more measurements and if the behavior of the baseline repeats - just neglect it

Have you tried to use a measured of the original sample before The treatment and some pattern reference absorption bands? The same sample that you are applying The treatment. Such as a “zero time” , for example. And of course, you can do that by Follow another absorption band at spectra that doesnt changes with the evolution treatment. This Way you can use a relation and neglect with any problems because you are working with a reference absorption band. this method is very helpful to follow a reaction progress, for example.

Dear Leonid, Alaa,

thank you for your expert opinion, All FTIR measurements are taken by scanning the sample atleast 3 times and i would still get the similar trend.

Dear Ana,

Thank you for your response. I have colelcted data about the pristine sample at the moment and then this will be compared with the treated ones. There would be no problem in comparing the reference with the measured results, however, the method to create a baseline would be a problem, as explained earlier. I am adding an image below and hope this may represent a clear pricture of my question.

Dear Ahmed,

Thank you for your answers. I can see that it works with spectra that are pretty clean. However, even if you adjust the 'y' value using contant mode, it is impossible to cover all of the aspects without an error. Please see the image below. This is a made up example and NOT AN ACTUAL SPECTRA.

It can be seen that with constant mode, the baseline would cover areas that are close to this baseline with low %e, however, with areas or peaks that fall a considerable magnitude below the baseline will have a large error in their areas calculated. In contrast, the second derivative method would help in picking up points and tailoring the baseline as per needs.

My Question: is it a widely accepted method to create a baseline like i have created using second derivative method?

Thank you

Well, for starters I think it is important to make sure you are getting valid data to begin with before being concerned with the data workup. Ideally you would avoid spectral manipulation, especially manual manipulation, as these are difficult to reproduce from sample to sample. I don't know Origins software, but there are many packages for doing quantitative analysis (EFTIR is what I use). Second derivatives and related tools are used for looking at shoulders of peaks or otherwise "playing" with the data. I shouldn't think it appropriate for picking baselines.

Per your data, first you should work in Absorbance, not % transmittance. Absorbance is linear with concentration. %T will give you a log relationship.

Second, many valleys between peaks don't make it to the baseline. Broad peaks that overlap will never get to baseline regardless the resolution of your instrument. If you artificially force peaks to baseline, you may be distorting the actual peak width, and that distorts your results.

Third, you don't mention the type of sample method or technique. Some sampling methods give different peak heights and baselines. For example solid sampling per KBr pellet often has a rising baseline at higher wavenumbers, and ATR techniques will have less intense peaks at lower wavenumbers.

Good luck!

Dear Joe Van Gompel ,

Thank you for your valued input, i will have a look at EFTIR again. Origin was used only for plotting.

Regards