Multi-Sensor Approach for Enhanced Characterization of Lithium Ore

Global trends indicate that the demand for many Critical Raw Materials (CRMs) and Strategic Raw Materials (SRMs) is rising and is expected to increase dramatically in the near future. These minerals are essential for key industries, including automotive and electronics, and serve as crucial enablers of the green energy transition, playing a vital role in achieving net-zero climate targets. To meet this increasing demand, it is crucial to enhance the characterization and modelling of these materials to better understand their quantity and distribution in both primary and secondary resources, such as mine waste. Sensor technologies could provide an effective solution for raw material characterization, supporting this effort. In this work, a data-driven methodology employing machine learning techniques is proposed. It utilizes laser-induced breakdown spectroscopy (LIBS) and visible-near infrared/short-wave infrared (VNIR/SWIR) spectral data to achieve more accurate characterization of lithium, a critical SRM, in lithium-bearing pegmatite deposits. The methodology commences with data exploration and pre-processing, followed by an evaluation of the techniques' effectiveness in element and mineral identification. This is followed by data modelling and validation. The collected spectral data were used to develop classification models, using Support Vector Classification (SVC) and Linear Discriminant Analysis (LDA), as well as predictive models for the prediction of Lithium concentration using Partial Least Squares Regression (PLSR) and Support Vector Regression (SVR). The results show that using these techniques coupled with machine learning significantly enhances the compositional analysis of lithium ore. The findings suggest that this approach can improve material characterization, enable effective process control, and help define the requirements for mineral processing. As a result, it could potentially increase the efficiency of mining and re-mining operations.