Fluid-melt trace-element partitioning behaviour between evolved melts and aqueous fluids: Experimental constraints on the magmatic-hydrothermal transport of metals

Abstract The fluid-melt partition coefficients of ten trace metals (Li, Mn, Cu, Zn, Rb, Sr, Mo, W, and Pb, along with K) between hydrous evolved melts and coexisting Cl-bearing fluid(s) have been experimentally determined at 850–1050 °C, 0.6–2.2 kbar, and ƒ O 2  ≈ NNO –0.5 to NNO +2 log units. The measured compositions of the quenched fluid phases vary from low salinity (~2 wt% Cl − ) to >19 wt% Cl − , with maximum starting fluid concentrations of >21 wt% Cl. The results of 15 partitioning experiments confirm that D metal fluid/melt values depend strongly on the melt and fluid composition, with the greatest metal affinities for the fluid phase shown in the experiments with high-Cl fluids, and the most fractionated melts. Five experiments conducted in Au 96 Cu 4 capsule alloy show D Cu fluid/melt values ranging from ~40 to >80, and a strong correlation with melt FeO wt% contents at equivalent fluid Cl − contents. In contrast, Li shows only weak partitioning in favour of coexisting fluids across all melt compositions, with D Li fluid/melt ranging from 0.16 in low salinity fluids to 1.17 in the highest Cl fluid (19.15 wt% Cl − ). A procedure was employed to redissolve and analyse the aluminosilicate precipitates that form during quenching of the experiments. This procedure allows for metal contents of the final quench fluids to be constrained by both mass balance and direct measurement by solution ICP-MS analysis. A comparison of the mass balance calculated vs. directly measured fluid-melt partition coefficients generally gives good agreement, within an order of magnitude for most trace elements. However, discrepancies between these methods arise due to errors inherent in the mass balance calculations for metals with low abundances in the fluids (e.g. Sr), and especially those showing affinities to alloy with the experiment capsule material (e.g. Mo). These discrepancies in constraining experimental fluid trace element budgets also suggest that the precious metal capsule alloys cannot always be considered completely inert containers in experimental investigations, and semi-quantitative LA-ICP-MS analysis of the raw capsule alloys confirms the detectable presence of several of the trace elements investigated here. Integrating these new partitioning data with previous experimental and natural studies confirms that aqueous Cl-bearing fluids exsolved by hydrous calc-alkaline melts stored at mid- to upper-crustal conditions are capable of efficiently sequestering and transporting some economically important trace metals (e.g. Mn, Cu, Zn). However, the new fluid/melt Li partitioning data, in conjunction with evidence of very late (post-eruptive) Li modification, suggests that Li is moderately-strongly fluid-immobile at depth and suffers from significant post-eruptive re-equilibration. This has implications for deciphering the signatures of ‘volatile fluxing’ in arc magmas that are preserved in mineral and melt inclusion trace element concentrations, and the development of mineralised stockwork regions in porphyry-style metal deposits.