Near-infrared as a swift method to quantify dioctahedral smectite content in bentonite

Abstract

Near-infrared spectroscopy is one of the fastest available analytical methods, and it is susceptible to clay minerals in the studied samples or terrains, making it potentially valuable for quantifying clay minerals' content. Nevertheless, its practical applicability for this task needs to be rigorously tested. In this study, we sought to evaluate its effectiveness by retrieving the smectite quantity from controlled binary, ternary, and polymineralic mixtures with common bentonitic components based on the measured spectra using the linear spectral unmixing technique. Encouragingly, the results were auspicious, with an average deviation of (15 % for quartz, and ∼ 7 % for feldspar) binary mixtures, ∼7 % for ternary mixtures, and ∼ 3 % of the actual values for polymineralic mixtures. Combining multiple parameters from different absorption features in the unmixing model provided greater statistical independence. These parameters were selected based on their statistical significance. Stepwise regression indicated that the most dependent parameters regarding smectite content are related to the 2.20 μm (4530 cm⁻¹) absorption feature, these findings support the previous studies which utilized the OH vibration region for smectite quantification. The developed automated unmixing model offers several benefits for quantification: it is fast, straightforward, and neither requires complex sample preparation nor the development of a comprehensive standard library or dataset.