# Expected accuracy of experimental IR spectra compared to literature reference

I have a table from a journal which indicated signals of an alpha chitin that I used as theoretical values. I did my own analysis to identify a chitin sample and it had signals that slightly varied from the theoretical (ex. $3106\ \mathrm{cm}^{-1}$ symmetrical N-H vibration and I got $3103\ \mathrm{cm}^{-1}$). Should I consider this? And what caused the variation?

• seems pretty close to me, only 3 in 3100; what is the resolution with which the spectra was taken, are the lines overlapping , was the sample environment the same etc? These may explain the very small difference as may slight changes in calibration. May 3 '17 at 14:14
• Hello @porphyrin, the lines weren't overlapping and the sample environment were the same.
– Byte
May 4 '17 at 14:01
• It is a very good agreement, most likely within experimental error. May 4 '17 at 15:27

In addition to @porphyrin's comment, the interaction of your sample with the surrounding may overall cause some changes in shape and place of the absorption bands; namely if your analysis

• follows the classical sample preparation, grinding your compound altogether with excess of KCl to press a pellet (ionic environment) / mixed in nujol (apolar environment) that is then characterised in transmission

.or.

• is an analysis by ATR with single or multiple reflection where your sample basically is a neat film on top of your optical window; either probe head or between crystal and anvil. (The ATR correction offered by the spectrometer's software, not every time active, basically will account for the penetration depth varying by the wavelength used, hence influencing rather the displayed %transmittance, then the position of the barycentre of the absorption band in question.)

If you aim for a qualitative IR analysis of a small organic molecule, I would not worry too much if the reported IR band was reported at 3106 cm$^{-1}$ and is now spot at 3103 cm$^{-1}$. The iS10 ATR-FT-IR spectrometer by Nicolet/ThermoScientific, for example, provides a data spacing of around 0.7 cm$^{-1}$. Performing a library search in an electronic database to identify / narrow which compound the sample in question is hopefully would not be affected, either.

(A quantification, like determining the composition of gasoline by PCA, would of course need some calibration.)

The original question however may be read as if you aim to monitor the (partial) digestion of larger proteins and polysaccharides, to say something about their secondary structures. A shift of 10...20 cm$^{-1}$ of an amide absorption band may be relevant; yet then more often the ones around 1600...1700 cm$^{-1}$ are taken into consideration.

• Hello @buttonwood, I used a ThermoScientific FT-IR spectrometer with an old OMNIC software and proceeded with the KBr pellet method. I was aiming for a qualitative analysis and got confused with the slight difference. Thanks for clarifying it though!
– Byte
May 4 '17 at 14:03