TY - JOUR
T1 - Montmorillonite content prediction in bentonite using Vis–NIR spectroscopy and PLSR analysis
T2 - Effects of humidity and mineralogical variability
AU - Seo, Chanyoung
AU - Jo, Ho Young
AU - Byun, Yujin
AU - Ryu, Ji Hun
AU - Joo, Yongsung
N1 - Publisher Copyright:
© 2024 The Authors
PY - 2024/8
Y1 - 2024/8
N2 - Bentonite, mainly composed of montmorillonite, has unique physicochemical properties, such as a high swelling capacity, low hydraulic conductivity, and high cation exchange capacity. The properties of bentonite significantly depend on its montmorillonite content, making quantifying montmorillonite essential for evaluating bentonite. Traditional methods such as X-ray diffraction analysis often encounter difficulties due to the structural and elemental variability of clay minerals. In contrast, spectroscopy can provide a fast and cost-effective alternative with the benefits of straightforward preprocessing and measurement. This study aimed to develop calibration models for predicting montmorillonite content in bentonite using visible and near-infrared (Vis–NIR) spectral features combined with partial least squares regression (PLSR) analysis. Quartz, feldspar, Ca-bentonite (KCa-B), and Na-bentonite (GNa-B) were used in this study. Montmorillonites (KCa-M and GNa-M) were extracted from their respective bentonites. Binary and ternary mixtures of these minerals were then prepared and analyzed spectrally in the 350–2500 nm range. Correlations between montmorillonite content and spectral features were derived using PLSR, with evaluation via the leave-one-out cross-validation method. The resulting model demonstrated high accuracy with R2 and RMSE values of 0.917 and 8.6 wt% for Ca-montmorillonite and 0.936 and 7.5 wt% for Na-montmorillonites, respectively. Independent validation confirmed the effectiveness of the model. Furthermore, adjustments for humidity based on Vis–NIR spectral variations can potentially enhance the precision of the prediction. The study highlights the potential of Vis-NIR spectroscopy as a reliable tool for predicting montmorillonite content in bentonite.
AB - Bentonite, mainly composed of montmorillonite, has unique physicochemical properties, such as a high swelling capacity, low hydraulic conductivity, and high cation exchange capacity. The properties of bentonite significantly depend on its montmorillonite content, making quantifying montmorillonite essential for evaluating bentonite. Traditional methods such as X-ray diffraction analysis often encounter difficulties due to the structural and elemental variability of clay minerals. In contrast, spectroscopy can provide a fast and cost-effective alternative with the benefits of straightforward preprocessing and measurement. This study aimed to develop calibration models for predicting montmorillonite content in bentonite using visible and near-infrared (Vis–NIR) spectral features combined with partial least squares regression (PLSR) analysis. Quartz, feldspar, Ca-bentonite (KCa-B), and Na-bentonite (GNa-B) were used in this study. Montmorillonites (KCa-M and GNa-M) were extracted from their respective bentonites. Binary and ternary mixtures of these minerals were then prepared and analyzed spectrally in the 350–2500 nm range. Correlations between montmorillonite content and spectral features were derived using PLSR, with evaluation via the leave-one-out cross-validation method. The resulting model demonstrated high accuracy with R2 and RMSE values of 0.917 and 8.6 wt% for Ca-montmorillonite and 0.936 and 7.5 wt% for Na-montmorillonites, respectively. Independent validation confirmed the effectiveness of the model. Furthermore, adjustments for humidity based on Vis–NIR spectral variations can potentially enhance the precision of the prediction. The study highlights the potential of Vis-NIR spectroscopy as a reliable tool for predicting montmorillonite content in bentonite.
KW - Humidity
KW - Montmorillonite
KW - partial least squares regression (PLSR)
KW - Quantitative analysis
KW - visible and near-infrared (Vis–NIR) spectra
UR - http://www.scopus.com/inward/record.url?scp=85199463233&partnerID=8YFLogxK
U2 - 10.1016/j.geoderma.2024.116980
DO - 10.1016/j.geoderma.2024.116980
M3 - Article
AN - SCOPUS:85199463233
SN - 0016-7061
VL - 448
JO - Geoderma
JF - Geoderma
M1 - 116980
ER -