The origin of halide melt phases in layered intrusions, and their significance to platinum-group element mobility

Fluid and melt inclusions are preserved within pegmatite bodies and cumulus minerals within mafic-ultramafic layered intrusions that host economic concentrations of the platinum-group elements (e.g., Bushveld Complex, South Africa; Stillwater Complex, Montana). The inclusions indicate that the earliest volatile phase to have exsolved from the crystallizing intrusions was a relatively anhydrous carbonic fluid (CO2-dominated). As crystallization proceeded, volatiles became increasingly water-rich and saline, consistent with the relative saturation limits of carbonic and aqueous fluids in mafic silicate liquids, and the partitioning behavior of Cl in fluid-melt systems. Previously unreported, the latest stage volatiles in the layered intrusions were halide melts (slightly hydrous molten salts) of relatively simply composition (NaCl with minor KCl or CaCl2) with salinities in excess of 90 wt% eq. NaCl or CaCl2. These volatiles were trapped at minimum temperatures of 760-800°C, near the eutectic temperature for water-saturated granitic liquid at moderate crustal pressures. Trace element analysis of the salt melt inclusions by laser ablation ICP-MS (ETH Zürich) show that they contain no detectable concentrations of ore and accessory metals. This is in contrast to the earlier, lower salinity volatiles which contain ppm-concentrations of Pt, Pd, As, Bi, Sb as well as abundant S and base metals. Heterogeneous entrapment of late-stage silicate melt and halide melt provides unambiguous evidence for the coexistence of both phases. However, experimental constraints on the nature of exsolved volatiles from mafic or felsic silicate liquids suggest that the halide melt phases cannot represent an exsolved phase from that coexisting silicate liquid, since this would require unrealistically high (initial) Cl:H2O ratios for the parental silicate liquid (> 9 for a granitic residue). Analysis of rhyodacitic silicate melt inclusions that coexist with the halide melt inclusions show that the coeval silicate melts had Cl:H2O ratios of only 0.1 to 0.2. Similarily, the salt melt phases could not have evolved via the removal of H2O by crystallization of hydrous magmatic minerals (e.g., biotite, apatite) since their modal abundances in the intrusions are very low. The most plausible explanation for the halide melt phases involves the "dehydration" of an initially lower- salinity aqueous fluid. This may have occurred by the reaction of the aqueous fluid with nominally-anhydrous minerals such as pyroxene, or by the late-stage alteration of cumulus minerals to hydrous mineral assemblages. Through the use of conventional hydrothermal experimental techniques, it can be shown that the reaction of a volumetrically-minor CaCl2-rich aqueous fluid phase (20 wt% eq. CaCl2) with the assemblage diopside-enstatite-quartz at near-solidus conditions (700°C, 0.4 kbar) results in the formation of tremolite by the reaction of H2O with the initially anhydrous mafic mineral assemblage. The resulting salinity of the dehydrated saline phase, trapped as synthetic inclusions in quartz, was > 96 wt% eq. CaCl2, consistent with the water-poor nature of the salt melt inclusions from the intrusions. The results of this study indicate that, through the loss of H2O, metal-bearing aqueous volatiles in layered intrusions may precipitate metals as they are dehydrated to form salt melt phases. Metal precipitation may occur as amount of free H2O in the volatile phase necessary to hydrate metal complexes decreases. This precipitation mechanism challenges the conventional magmatic hypothesis for platinum-group element deposit formation in layered intrusions.