Extraction of sulfides from the partially molten mantle is vital to elucidate the cycling of metal and sulfur elements between different geochemical circles but has not been investigated systematically. Using laboratory experiments and theoretical calculations, this study documents systematical variations in lithologies and compositions of silicate minerals and melts, which are approximately consistent with the results of the thermodynamically-constrained model. During a melt-peridotite reaction, the dissolution of olivine and precipitation of new orthopyroxene generate an orthopyroxene-rich layer between the melt source and peridotite. With increasing reaction degree, more melt is infiltrated into and reacts with upper peridotite, which potentially enhances the concomitant upward transport of dense sulfide droplets. Theoretical analyses suggest an energetically focused melt flow with a high velocity (~ 170.9 μm/h) around sulfide droplets through the pore throat. In this energic melt flow, we, for the first time, observed the mechanical coalescence of sulfide droplets, and the associated drag force was likely driving upward entrainment of fine μm-scale sulfide. For coarse sulfide droplets whose sizes are larger than the pore throat in the peridotite, their entrainment through narrow constrictions in crystal framework seems to be physically possible only when high-degree melt-peridotite reaction drives high porosity of peridotite and channelized melt flows with extremely high velocity. Hence, the melt-rock reaction could drive and enhance upward entrainment of μm- to mm-scale sulfide in the partially molten mantle, potentially contributing to the fertilization of the sub-continental lithospheric mantle and the endowment of metal-bearing sulfide for the formation of magmatic sulfide deposits.

The chemical composition of the deep continental crust is key to understanding the formation and evolution of the continental crust. However, constraining the chemical composition of deep continental crust is limited by indirect accessibility. Here we present a modeling method to constrain deep crustal chemical structures from observed crustal seismic structures. We first compile a set of published composition models for the continental crust. Phase equilibria and compressional wave speeds (VP) are calculated for each composition model at a range of pressure and temperature (278–2223 MPa, 50–1200°C). Functional relationships are obtained between calculated wave speeds and crustal compositions at pressure and temperature conditions within the alpha(α)-quartz stability field. These relationships can invert observed seismic wave speeds of the deep crust to chemical compositions in regions with given geotherms (MATLAB code is provided). We apply these relationships to wave speed constraints of typical tectonic settings of the global continental crust and the North China Craton. Our method predicts that the lower crust in regions with thin- (e.g., rifted margins, rifts, extensional settings, and forearcs) or thick-crust (e.g., contractional orogens) is more mafic than previously estimated. The difference is largest in extensional settings (52.47 ± 1.18 and 51.11 ± 1.61 vs. 59.37 wt. % SiO2). The obtained 2-D chemical structure of the North China Craton further shows features consistent with the regional tectonic evolution history and xenoliths. The obtained chemical structure can serve as a reference model from which chemical features in the deep crust can be recognized.