Application of FTIR-PAS and Raman spectroscopies for the determination of organic matter in farmland soils

Abstract

In soil analysis, Raman spectroscopy is not as widely used as infrared spectroscopy mainly owing to fluorescence interferences. This paper investigated the feasibility of Fourier-transform infrared photoacoustic (FTIR-PAS) and Raman spectroscopies for predicting soil organic matter (SOM) using partial least squares regression (PLSR) analysis. 194 farmland soil samples were collected and scanned with FTIR and Raman spectrometers in the spectral range of 4000-400cm(-1) and 180-3200cm(-1), respectively. For the PLSR models, the combined dataset was split into 146 samples as the calibration set (75%) and 48 samples as the validation set (25%). The optimal number of analytical factors was determined using a leave-one-out cross-validation. The results showed that SOM could be predicted using FTIR-PAS and Raman spectroscopies independently, with R(2)>0.70 and RPD>1.8 for the validation sets. In comparison to the single applications of FTIR-PAS and Raman spectroscopies, accurate prediction of SOM was made by combining FTIR-PAS and Raman spectroscopies, with R(2)=0.81 and RPD=2.18 for the validation sets. By statistically assessing large amounts of PLS models, model-population analysis confirmed that the accuracy of the PLS model can be increased by combining FTIR-PAS and Raman spectroscopies. In conclusion, the combination of FTIR-PAS and Raman spectroscopies is a promising alternative for soil characterization, especially for the prediction of SOM, owing to the availability of complementary information from both FTIR-PAS (polar vibrations) and Raman spectroscopy (non-polar vibrations).

Keywords: FTIR-PAS spectroscopy; Model-population analysis (MPA); Partial least-squares regression (PLSR); Principal component analysis (PCA); Raman spectroscopy; Soil organic matter.