Machine learning of lithium and gallium accumulation based on the major element contents in coal gangue from Shanxi province coalfields

Abstract

Coal gangue, a byproduct of coal mining and processing, contains valuable elements like aluminum, silicon, iron, calcium and titanium, along with a certain amount of lithium (Li) and gallium (Ga). Shanxi, a major coal-producing province in China, produces a substantial amount of coal gangue with Li and Ga accumulation, indicating potential for resource recycling. Nevertheless, the low concentration and dispersed distribution characteristics of Li and Ga in coal gangue pose challenges for their direct determination. This study employed machine learning to assess eight models for analyzing Li and Ga accumulation in coal gangue based on its major components. The results indicated that Li and Ga contents in coal gangue from Shanxi Province coalfields vary widely, with Li and Ga content ranging between 50-300 ppm and 20–40 ppm, respectively. The XGB model showed superior performance in analyzing Li and Ga contents in coal gangue, revealing that Al₂O₃, TiO₂ and loss on ignition (LOI) in coal gangue are positively correlated with Li and Ga contents; while K₂O and Na₂O are negatively correlated with Li content, and K₂O, Na₂O and Fe₂O₃ are negatively correlated with Ga content. The diverse correlations between the major components and the Li and Ga contents in coal gangue may be attributed to the charge imbalance structures of aluminosilicate minerals, isomorphic ion substitution in aluminosilicates, and the physical and chemical adsorption properties of organic matter. This study provides theoretical guidance for exploring the occurrence forms and high-value utilization of trace elements such as Li and Ga in coal gangue.