The genesis of the Mesozoic Au deposit within the Jiaodong Peninsula is debated, with vague connections to magmatism, deep-source metamorphic devolatilization, or the involvement of mantle materials over the long term. This study presents an integrated investigation involving textures, in-situ geochemical analysis, and machine learning discrimination of the rutile of the Xiadian Au deposit at the northwest Jiaodong Peninsula, aiming to provide metallogenic insights. Rutile is distinguished into three types: Type 1 rutile formed in the pre-ore stage is fine-grained and replaces and coexists with relict titanite, with euhedral crystal and being homogeneous and dark BSE response. Type 2 and type 3 rutile exhibit subhedral and anhedral crystals and are associated with quartz-sericite-pyrite alteration. The former has low W contents and is dark BSE response, while the latter is brighter in BSE and exhibits pronounced fragmentation and erosion. Chemical trends indicate the elemental substitution mechanism within rutiles from early to later stages: (Nb, Ta)5+ + (V, Fe, Al, Cr)3+ ↔ 2Ti4+ to 2(V, Fe, Al, Cr)3+ + W6+ ↔ 3Ti4+. Elevated concentrations of trivalent iron ions due to elevated oxygen fugacity of the ore-forming fluids are more likely to have contributed to this change, rather than the increment of W content. This may induce the destabilization of gold-sulfur complexes and contribute to gold mineralization. Machine learning models, including Extreme Randomized Tree (ET), Support Vector Machine, and Naive Bayes classifiers, were trained using a collected dataset containing 6598 rutile geochemistry data. These models consistently classify Xiadian rutile as the hydrothermal origin and having affinity to orogenic Au deposit. The SHapley Additive exPlanation (SHAP) framework was used to explain the discrimination criteria for the best-performing ET models. It revealed that elements such as Y, Ta, Mo, Sn, W, and Nb significantly contribute to classifying Xiadian rutile as affiliated with orogenic Au deposit. Further mass balance calculations with titanite, the predominant precursor Ti-bearing mineral of rutile, suggest that the W content in rutile is controlled by ore-forming fluids, likely having a deep source. In contrast, elements like Y, Mo, Cr, Nb, and Ta are influenced by leaching or re-equilibration with host rocks. Collectively, deep geodynamic processes likely provide major Au mineralization-related materials and fluids, while elemental leaching or re-equilibration with shallow crust rocks near fluid pathways influences the geochemical characteristics of the Xiadian gold system.
Read full abstract