The chilled marginal rocks of the Panzhihua layered intrusion of the Emeishan Large Igneous Province (ELIP), SW China, including microgabbros and olivine-phyric gabbros, are similar to coeval high-Ti basalts in the Panzhihua area in both major and trace element compositions. The olivine-phyric gabbros are characterized by depleted Nd isotopic compositions (epsilon(Nd)(t) = +115 to +418) and initial Sr-87/Sr-86 (07043-07052) similar to those of the Panzhihua layered series and the associated high-Ti basalts. The microgabbros have similar initial Sr-87/Sr-86 (07045 to 07054) but more enriched Nd isotopic compositions (epsilon(Nd)(t)=-149 to +006), suggesting that the magma was contaminated by lower crustal materials. The microgabbros have average compositions of 455wt % SiO2, 85wt % MgO and 135wt % FeOT, which are within the ranges of the associated high-Ti basalts. The average TiO2 content in the microgabbros is 27wt %, which is close the lower end of the high-Ti basalts (TiO2 = 25wt %). Coexisting olivine-phyric gabbros contain abundant coarse-grained olivine phenocrysts (20-60vol. %) and fine-grained clinopyroxene + plagioclase. Cr-spinel inclusions enclosed in the olivine phenocrysts have highly variable Cr2O3 contents (59-338wt %) within a single host crystal. Using the average composition of the chilled microgabbros as a starting composition, MELTS simulation indicates that such a magma can well reproduce the mineral compositions and assemblages in the overlying layered series of the Panzhihua intrusion, implying that the average composition of the chilled microgabbros is similar to the parental magma for these rocks. The chilled marginal rocks (microgabbros and olivine-phyric gabbros) have high S contents (up to 148wt %), but low abundances of platinum group elements (PGE). The total amounts of PGE in these rocks vary from 86 to 329ppb. The Cu/Pd ratios of these rocks range between 075x10(4) and 344x10(4), which are significantly higher than mantle values and indicate previous sulfide removal from the magma at depth. The parental magma with elevated Cu/Pd contains more MgO (>8wt %) than the coeval PGE-undepleted basalts (MgO = 32-45wt %), indicating that sulfide saturation was not triggered by extensive fractionation. Nevertheless, the mantle-like Sr-Nd-O-S isotopic compositions of the mineralized Fe-Ti oxide-bearing intrusions of the ELIP also suggest that S saturation was not related to crustal contamination. We suggest that the contamination and fractionation of an early pulse of mantle-derived S-undersaturated magma in a deep-seated magma chamber most likely caused S saturation and sulfide liquid segregation within the lower crust. The chilled microgabbro formed by influx of the residual PGE-depleted, but Nd isotope enriched, magma. A subsequent pulse of S-undersaturated magma then reached S saturation during the early stages of differentiation as a result of the resorption of the previously formed sulfide liquid.

This Nd isotope and PGE-depleted magma was then successively emplaced into a shallow crustal chamber along with captured olivine, generating the olivine-phyric gabbros and associated Fe-Ti oxide deposits. Our results support the interpretation that the Fe-Ti oxides directly crystallized from a basaltic parental magma at an early stage of differentiation and then formed the ore layers by gravitational settling. The composition of the parental magma does not support the hypothesis that the early crystallization of Fe-Ti oxides was controlled by high Fe and Ti concentrations in the parental magma. The moderately high oxygen fugacity (FMQ+1 approximate to FMQ+25) of the parental magma, which was inherited from an oxidized mantle source, may account for the early saturation of Fe-Ti oxides.