Rare earth element ore deposits associated with carbonatite derived from Earth’s mantle supply half of the world’s rare earth element. However, the formation of carbonatite and initial enrichment and transport of rare earth element in the mantle, is unclear. Here, we image the lithospheric architecture of a Cenozoic rare earth element ore belt in southwestern China by integrating seismic tomography with geochemical data. The subduction of the Indian continent caused vertical upwelling and lateral flow of the asthenosphere, which triggered the melting of the overlying subcontinental lithospheric mantle to generate carbonatites. Such a mantle source that previously metasomatized by fluids from recycled marine sediments is a precursor process critical for forming a giant rare earth element system. For the studied ore belt, three key factors are prerequisites to generating ore-forming carbonatites: thick lithosphere with a continental root; prior fertilization of the subcontinental lithospheric mantle; and trans-lithospheric weakness for magma ascent.

The sources of fluids and metals in porphyry systems of continental-collision settings are poorly constrained. Mercury isotopes display unique mass-independent fractionation (expressed as Δ ¹⁹⁹ Hg) and may provide important constraints on metal and volatile sources given that Hg is a highly volatile metal. Here, we report Hg isotope data on ore-forming porphyries, barren magmatic rocks, and mantle-derived mafic magmas from southern Tibet. The fertile porphyries and coeval mafic magmas display mainly positive Δ ¹⁹⁹ Hg values (up to +0.25 per mil), while Δ ¹⁹⁹ Hg values in barren magmatic rocks and mafic magmas are largely negative (−0.54 to 0.00 per mil). The positive Δ ¹⁹⁹ Hg values observed here are consistent with seawater and marine sediments, suggesting that the ultimate source of fluids involved in the genesis of post-subduction porphyry copper deposits was the mantle lithosphere metasomatized by previous oceanic plate subduction. Our Hg isotope data provide an alternative view to current metallogenetic models on collisional porphyry systems that focus on melting of the lower continental crust.

The generation of strongly potassic melts in the mantle requires the presence of phlogopite in the melting assemblage, while isotopic and trace element analyses of ultrapotassic rocks frequently indicate the involvement of subducted crustal lithologies in the source. However, phlogopite-free experiments that focus on melting of sedimentary rocks and subsequent hybridization with mantle rocks at pressures of 1–3 GPa have not successfully produced melts with K₂O >5 wt%–6 wt%, while ultrapotassic igneous rocks reach up to 12 wt% K₂O. Accordingly, a two-stage process that enriches K₂O and increases K/Na in intermediary assemblages in the source prior to ultrapotassic magmatism seems likely. Here, we simulate this two-stage formation of ultrapotassic magmas using an experimental approach that involves re-melting of parts of an experimental product in a second experiment. In the first stage, reaction experiments containing layered sediment and dunite produced a modally metasomatized reaction zone at the border of a depleted peridotite. For the second-stage experiment, the metasomatized dunite was separated from the residue of the sedimentary rock and transferred to a smaller capsule, and melts were produced with 8 wt%–8.5 wt% K₂O and K/Na of 6–7. This is the first time that extremely K-enriched ultrapotassic melts have been generated experimentally from sediments at low pressure applicable to a post-collisional setting.

Sulphur is the third most abundant volatile element in deep Earth systems. Analytical methods for accurately and efficiently determining the sulphur content and oxidation state in natural minerals are still lacking. Natural apatite is widely distributed in the Earth and incorporates a large amount of sulphur. Therefore, apatite is an ideal mineral for performing sulphur measurements. Here, we used spectroscopic, Raman, X-ray diffraction, laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS), electron microprobe (EMPA) and micro-X-ray fluorescence spectrometry (micro-XRF) analysis techniques and developed a new analytical approach (i.e., micro-X-ray absorption near-edge structure (micro-XANES) analysis of the sulphur K-edge) to investigate the chemical characteristics of natural apatite. These multiple methods were developed to measure in situ sulphur concentration and S oxidation states and to assess a potential natural apatite reference material. Apatite contains chemically homogeneous sulphur, with micro-XANES located at the peak energies corresponding to S⁶⁺ (sulphate; ~2482 eV), S⁴⁺ (sulfite; ~2478 eV), and S²⁻ (sulphide; ~2467, 2470 and 2474 eV). The Durango apatite contains total S presented as SO₃ at amount of 0.332 ± 0.012 wt.% (1σ), with a large amount of S⁶⁺ and a small contribution of S⁴⁺. The Kovdor apatite contains 44–100 ppm of S and is dominated by S⁶⁺. These results indicate that the Durango apatite crystallised under relative oxidising conditions, and the Kovdor apatite has a higher oxygen fugacity than Durango. In addition, this study indicates the potential use of the natural apatite reference material with its S composition and S oxidation state.

Apatite major and trace element chemistry is a widely used tracer of mineralization as it sensitively records the characteristics of the magmatic-hydrothermal system at the time of its crystallization. Previous studies have proposed useful indicators and binary discrimination diagrams to distinguish between apatites from mineralized and unmineralized rocks; however, their efficiency has been found to be somewhat limited in other systems and larger-scale data sets. This work applied a machine learning (ML) method to classify the chemical compositions of apatites from both fertile and barren rocks, aiming to help determine the mineralization potential of an unknown system. Approximately 13 328 apatite compositional analyses were compiled and labeled from 241 locations in 27 countries worldwide, and three apatite geochemical data sets were established for XGBoost ML model training. The classification results suggest that the developed models (accuracy: 0.851–0.992; F1 score: 0.839–0.993) are much more accurate and efficient than conventional methods (accuracy: 0.242–0.553). Feature importance analysis of the models demonstrates that Cl, F, S, V, Sr/Y, V/Y, Eu*, (La/Yb)N, and La/Sm are important variables in apatite that discriminate fertile and barren host rocks and indicates that V/Y and Cl/F ratios and the S content, in particular, are crucial parameters to discriminating metal enrichment and mineralization potential. This study suggests that ML is a robust tool for processing high-dimensional geochemical data and presents a novel approach that can be applied to mineral exploration.