Removal of the ambient air features from fourier-Transform near-Infrared spectra

Presence of ambient air in the optical path of Fourier-Transform single-beam near-infrared spectrometers introduces features that are external to the sample of interest and deteriorate accuracy on the time scale that is, as our data shows, similar to that of spectral acquisition. This necessitates corrective action per -formed either by controlling the atmosphere within the instrument or by post-processing the spectrum. Since controlling the atmosphere introduces a variety of technical issues a vast majority of spectrome-ters are equipped with software tools for post-processing the spectrum. A noticeable absence of applied studies on compensation of atmospheric absorption in the near-infrared spectral region prompted us to propose such a procedure from the aspect of recent advances in modeling of the ambient air absorption in this region that, in combination with a comprehensive survey of various baseline-correction algorithms, yielded a procedure for calibrated subtraction of the background single-channel spectrum that accurately models the corrupting features of the ambient air. Comparisons between different models of the total absorption of ambient air showed significant improvements that were characterized collectively, using multivariate analysis, and individually, using our novel de-trending technique, called self-subtraction. We conclude that corrections introduce measurable changes in the spectra and that implementation of this, or similar, algorithms is a prerequisite for performing quantitative analyses in the near-infrared region.(c) 2023 Elsevier Ltd. All rights reserved.

Automated summary

The presence of ambient air in Fourier-Transform single-beam near-infrared spectrometers introduces external features that affect the accuracy of spectral acquisition. Controlling the atmosphere or post-processing the spectrum is necessary to correct for these effects. We proposed a procedure that accurately models the corrupting features of the ambient air using advanced modeling techniques and comprehensive survey of baseline-correction algorithms. Comparisons between different models showed significant improvements in spectral analysis. Implementing these algorithms is essential for quantitative analysis in the near-infrared region.