Quantitatively characterization of rare earth ore by terahertz time-domain spectroscopy

Abstract

The Bayan Obo deposit is the world’s largest polymetallic associated minerals deposit of rare earths, iron and niobium, and the rarity of its physical properties restrict the knowledge and understanding of its laws. Taking the high-grade mixed rare earth concentrate of Bayan Obo as the research object, terahertz time-domain spectroscopy (THz-TDS) has been adopted for the systematic investigation of high-grade rare earth concentrate base on the traditional X-ray fluorescence (XRF), X-ray diffraction (XRD), scanning electron microscopy (SEM) and thermogravimetric-differential thermal analysis (TG-DTA). The absorption coefficient and refractive index of high-grade rare earth ores and their associated minerals of fluorite and dolomite, are all investigated by terahertz time-domain spectroscopy. The terahertz spectral response is affected by the type of mineral and its content. The acquired rare earth terahertz spectral data are processed by correlation analysis. Three machine learning algorithms, Partial Least Squares Regression (PLSR), Random forest (RF) and Multilayer Perceptron (MLP), are used to achieve quantitative detection of their concentrations and components with the coefficient of determination R² of the absorption coefficient of the optical parameter reaching up to 0.975, 0.992 and 0.984, respectively. This work promotes the growing understanding of terahertz transmission spectroscopy of rare earth-bearing minerals, which can be used to help guide the search for minerals, and to detect, identify as well as quantify them in geology. Terahertz time-domain spectroscopy supplies a new method for study of rare earth resources, and the comprehensive development and utilization of resources in the Bayan Obo deposit.