I'm analyzing some carbon nanotubes with DRIFTS. I mix the CNT with KBr powder and proceed to measure the spectrum of the powder. After acquiring the KBr background (in %T, peaks facing down) I proceed and acquire the sample spectrum and it comes back with the peaks in the opposite direction (up). Once you apply Kubelka-Munk Kubelka-Munk correction you get the spectrum as in transmission again but sometimes the KBr background is not completely removed. Why is this?.

  • $\begingroup$ Please clarify your question and show the spectra. The question heading is different but you are also talking about background correction. Whenever you do something new, always run a simple sample (just a simple polystyrene powder), whose results are known. This gives confidence in the results. Carbon FTIRs are horrible anyway. $\endgroup$
    – AChem
    May 21 at 20:11
  • $\begingroup$ Eons ago, I used direct transmission mode by employing a small amount of carbon in KBr rather this complicated Kubelka-Munk correction. Use a large number of scans, say 64. $\endgroup$
    – AChem
    May 21 at 20:12
  • $\begingroup$ KM theory, is more valid for weakly absorbing materials. Carbon absorbs everything...recall a black body is a good absorber. $\endgroup$
    – AChem
    May 21 at 20:16
  • $\begingroup$ @AChem I mixed the CNT with KBr and ran 64 scans. I also acquired a spectrum of Polystyrene and it was fine. The issue I describe seems to be sample related; also the KM correction is done by the software in like a second, I don't understand your comment. $\endgroup$ May 21 at 20:28
  • $\begingroup$ Is the sample and KBr bone-dry? $\endgroup$
    – AChem
    May 21 at 20:30

1 Answer 1


First of all I would suggest you to read the reference Validity limits of Kubelka–Munk theory for DRIFT spectra of photodegraded solid wood. Wood Sci Technol 45, 135–146 (2011). https://doi.org/10.1007/s00226-010-0314-x. The introduction highlights the how many spurious (or more crudely bogus) DRIFT results are out there just because researchers did not carefully think about the validity limits of KM.

The authors highlight the facts that KM is useful for weak absorbers. Certainly carbon nanotubes are not weak absorbers and one would expect problems (non-ideal behavior).

Although your question has been appreciated but it is missing a lot of information about the sample, its pre-treatment and many other critical instrumental settings. FTIR is a Fourier transform technique, and you have to choose (perhaps in advanced settings), the correct apodization function. Using an improper setting may cause final spectrum's artifacts. Check the manual which one is the right apodization for DRIFT. Make sure all the settings are correct from the manual.

Alternatively, a short-cut is that you repeat a well-behaving organic compound and reproduce a correct absorption spectrum which matches very well with a standard database and does not generate negative scales on the y-axis. You are also using 64 scans, which is fine.

The red-flag in that carbon spectrum is the negative y-axis. When the instrument generates an output which has no physical meaning...there is a problem somewhere. Even with simple double-beam UV-Vis spectrophotometers, some time the sample shows negative absorbance. It does not physically mean anything.

To troubleshoot one can re-visit the problem from the beginning.

(i) KBr is bone dry, FTIR grade and freshly dried.

(ii) The carbon nanotubes have been thoroughly washed with suitable high quality solvents and thoroughly dried > 24 hours. This will ensure the sample is clean and the background is not coming from random junk sticking to carbons. Carbon is an excellent non-selective adsorbent.

(iii) Mix the KBr and carbon nanotubes throughly (all IRs used to have mortar and pestle) so that the mixture is as uniform as possible.

(iv) After collecting background, use KBr as a sample and see how the baseline looks, and if this is reproducible or not. If it is reproducible, one could subtract this baseline from the final spectrum.

(v) Do a sample concentration study: Use different amounts of carbons in KBr (do it semi-quantitatively) and see how the spectrum changes. Overlay the spectra and see which regions of the spectrum change and when do they start to behave well. Carefully, use the same amount of the sample for DRIFT measurements.

(vi) In the worst case, if nothing works in DRIFT mode, use a simple KBr pellet method and use the transmission mode. A low sample concentration is required. Making a good pellet under high-pressure was an art in itself.

This is a little bit of work, but the end-results will be satisfying and you will be confident about the results.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.