Quantitative interpretation of results obtained from Attenuated total reflection Fourier transform infrared (ATR FTIR) spectroscopy is difficult and for deeper insight it is necessary to employ various data-processing methods. These methods must be suitable for handling large multidimensional data sets and for exploring the complete spectral information simultaneously. The effective implementation of these multivariate data analysis methods, however, also requires the pre-treatment of data. The pre-processing of raw data helps in the elimination of noise and the enhancement of discriminating features. This study focuses on two commonly-used multivariate methods of analysis: principal component regression (PCR) and partial least squares regression (PLSR); these methods enable the extraction of mineralogical information from infrared spectra. The present study also discusses the various spectral preprocessing methods that are widely used in ATR FTIR spectroscopy. A dataset of natural standards of common rock-forming minerals (calcite, dolomite, quartz, feldspar, muscovite, illite, smectite and kaolinite) and their synthetic mixtures was constructed to build PCR and PLSR models that link the mineralogy of the samples to their respective infrared spectral signatures. Infrared spectra of the samples were recorded from 400 to 4000 cm–1. As a reference, modal composition was also estimated from X-ray diffraction data. The resulting PCR and PLSR models were also tested on synthetic mixtures. The overall conclusion for the constructed 24 models is that, with respect to prediction, PCR and PLSR provide similar results. Different types of spectral treatment have greater impact on the estimated modal composition than the studied multivariate methods. Furthermore, in the models the respective amounts of various minerals were estimated with different uncertainties; this was the result of the difference in the infrared light-absorbing capacity of minerals, overlapping bands and other physical effects.
|Translated title of the contribution||Attenuated total reflection infrared spectroscopy combined with multivariate data analysis for studying modal composition|
|Number of pages||17|
|Publication status||Published - Jun 13 2018|
ASJC Scopus subject areas
- Geochemistry and Petrology