Understanding the geochemical behavior of elements during magmatism, weathering, and sedimentation is crucial for revealing their enrichment mechanisms. The Late Permian Xuanwei Formation claystones in southwestern China are enriched in rare earth elements (REEs), Y, Sc, Ga, Nb, and Zr. Using principal component analysis (PCA) combined with interpretable machine learning algorithms including random forest and eXtreme Gradient Boosting, we decipher their enrichment pathways. PCA reveals that REEs and Y, together with Ga, were co-enriched with high field strength elements (HFSEs) in late-stage accessory minerals during basaltic magmatism, whereas Sc was preferentially associated with early-forming mafic minerals. During sedimentation, REEs and Y were mobilized and enriched, whereas HFSEs remained largely immobile. Machine learning models identify key predictors controlling elemental enrichment, including Y, Sr, P, U, and Ga for REEs; REEs, Ga, and Hf for Y; Ti, V, and Fe for Sc; Th, REEs, and Y for Ga; and Ta, Hf, Th, and U for Nb and Zr. Based on these results, we propose a geochemical model for the enrichment of critical metals. During the magmatic stage, REEs, Y, and Ga were associated with HFSEs, whereas Sc was coupled with Fe–Ti–V-bearing mafic minerals. Weathering mobilized REEs and Y, while Sc, Nb, and Zr remained relatively inert. During sedimentation, REEs and Y became re-coupled with P- and Sr-rich phases, Sc shifted toward Fe–Ti associations, Ga became linked to Al-rich phases, and Nb and Zr continued to exhibit immobile behavior. Overall, enrichment of REEs and Y reflects the combined effects of magmatic inheritance, weathering mobilization, and sedimentary re-concentration, whereas enrichment of Ga and HFSEs is mainly inherited from magmatic minerals. These findings are consistent with mineralogical constraints, demonstrating the effectiveness of integrating geochemistry with multivariate statistics and machine learning to unravel element behavior across geological processes.
扫码关注我们
求助内容:
应助结果提醒方式:
