Geochemistry of muscovite and biotite in granites from tungsten deposits: a comparative study between Nanling and Jiangnan belts

• LA-ICP-MS analyses reveal that Li, V, Cr, and Mn are enriched in biotite, while Be, Ga, and Sn are enriched in muscovite within the same deposit. • The mica composition, along with temperature and oxygen fugacity data, suggests that the NMB tungsten deposits were influenced by a great volume of fluids compared to the JOB deposits. • A machine learning model based on mica compositions achieves an accuracy of 80.58 % in distinguishing granite ore belt and 100 % in identifying tungsten mineralization. The Jiangnan Orogenic Belt (JOB) and Nanling Mineralization Belt (NMB) are prominent tungsten-rich regions in South China, with significant tungsten mineralization associated with granite. This study investigates the granite-related tungsten mineralization process using geochemistry of mica, a critical metal carrier, from granites in the Dahutang, Xianglushan, Chuankou, and Xintianling deposits within these belts. Our results reveal that the biotite and muscovite from the JOB are depleted in Nb and Ta compared to those from the NMB, and the biotite from the JOB is more enriched in Rb compared to those from the NMB. Critical metals such as Li, V, Cr, and Mn are most likely to be hosted by biotite, whereas muscovite tends to enrich Be, Ga, and Sn. The data suggest that with decreasing K/Rb ratio, elements such as Li, Rb, Nb, Ta, Zr, and W in muscovite increase, indicating progressive enrichment of ore-forming elements during magmatic evolution. Biotite provenance discrimination diagram indicates that the ore-forming materials of the Dahutang and Xianglushan deposits are crust-derived, while the Xintianling material is from a mixed crust-mantle source. The higher Nb/Ta ratios and the crystallization temperature close to the solidus temperature in Xintianling biotite suggest significant fluid or fluid–melt interactions during granite formation. These differences highlight the role of host rock provenance and fluid or fluid–melt interactions in critical metals variations in mica between JOB and NMB. Furthermore, the results indicate that the K/Rb ratio and Nb/Ta ratio are valuable indicators for distinguishing between deposits from the JOB and NMB. The correlation of mica composition with tungsten mineralization suggests that tungsten enrichment is closely related to magmatic-hydrothermal evolution, particularly during fluid-induced alteration. Machine learning applications on geochemical data from mica provide an effective method for distinguishing different metallogenic belts (80.58 %) and tungsten mineralization (100 %), demonstrating the potential for using mica composition as a reliable tool for ore deposit classification and exploration. This study enhances the understanding of the magmatic processes and fluid interactions governing tungsten mineralization and provides valuable insights into the geochemical signatures of tungsten-bearing granites in South China. The findings also underscore the potential of mica composition for use in distinguishing metallogenic belts and predicting tungsten mineralization.