Assessment of estuarine sediment and sedimentary organic matter properties by infrared reflectance spectroscopy

The goal of this work was to evaluate the capability of mid-infrared diffuse reflectance (DRIFT) spectroscopy, coupled with multivariate chemometric analysis, to rapidly provide valuable information on sediment composition and organic geochemistry. A large (150) and heterogeneous set of estuarine sediments was analyzed in their natural matrix by DRIFT. Principal component analysis (PCA) of DRIFT spectra clearly discriminated sediments from distinct deposition events, such as landslides or floods, based on their higher carbonate contents. PCA also distinguished mostly terrigenous sediments, enriched in organic matter, aromatics and SiO2, from mostly marine sediments, enriched in clays and carbonates. Partial least-squares (PLS) regression of the DRIFT spectra was used to predict various sediment properties, without the extractions or alterations required by traditional analytical methods. The relative errors in quantifying properties, as expressed by the relative root-mean-squared error (%RMSE), were ~ 6% for total humic substance and humin content, 7.6% for %C, 11.5% for atomic N:C ratio, and 13–14% for %N, %C as humic substances and %C as humin. The study revealed the importance of using a calibration sample set that is representative of the samples to be quantified. This calibration can then be used for the future and rapid analyses of unknown samples. Geochemical processes could also be inferred from band assignments of the loadings of the PLS models. These loadings indicated that nitrogen, humin (%C as humin), and aluminosilicate contents were correlated, providing independent evidence that nitrogen-rich compounds are important bonding agents between the organic and mineral fractions of the sediments. DRIFT-PCA and DRIFT-PLS represent valuable techniques for a rapid and cost-effective screening of sediment core characteristics and for an evaluation of sediment complex composition.