QUANTIFICATION OF BRUCITE AND SERPENTINE IN PERIDOTITE ROCKS WITH UV/VIS/NIR SPECTROSCOPY AND MACHINE LEARNING MODELING

Brucite rich serpentinized peridotite (BRSP) rocks can be used as the storage for greenhouse gas as Brucite is known to react with carbon dioxide (CO2). Therefore, it is important to characterize the brucite content in the BRSP rocks before they can be applied for the carbon capture and storage (CCS) projects. Additionally, characterizing brucite from serpentine in these deposits can be challenging, resulting in a need for more precise technology.

UV/Vis/NIR spectroscopy has been proven as a promising tool in characterizing minerals. In this study, a full range (350nm – 2500nm) ultra high-resolution spectrometer is used to measure the UV/Vis/NIR spectra of brucite and serpentine rich rock powders. The full range spectral analysis allows for the identification and quantification of brucite and serpentine minerals with ultra high spectral resolution.

A case study was completed to build a Partial Least Square Regression (PLSR) model to quantify the brucite concentration in unknown samples. The calibration set includes data from 37 calibration samples and the test set includes 8 unknown samples. The regression model has high quality measures such as root mean square errors (RESMC = 0.0018, RMSECV = 0.0038, RMSEP) as well as determination coefficient (R² = 0.994). The predicted brucite/serpentine concentration is consistent with provided Thermogravimetric Analysis (TGA) results as a means for validation.