Petrologic and geochemical constraints on the petrogenesis of Permian–Triassic Emeishan flood basalts in southwestern China

The Emeishan flood basalt is a large igneous province erupted during the Permian–Triassic period in southwestern China. Based on petrographic, major and trace element, and Sr–Nd isotope data, the Emeishan basalts can be classified into two major magma types. These are: (1) a low-Ti (LT) type that exhibits low Ti/Y (500). The HT lavas can be further divided into three subtypes. HT1 lavas exhibit significantly high TiO2 (3.65–4.7%), Fe2O3* (12.7–16.4%), Nb/La (0.75–1.1), coupled with higher εNd(t) (1.1–4.8) and lower SiO2 (45–51%); HT2 lavas are compositionally similar to the HT1 lavas but show conspicuous depletion in U and Th. The HT3 type has higher Mg# (0.51–0.61) than the HT1 and HT2 lavas. It differs from the LT type in having higher TiO2 (∼3%) at comparable Mg#. Elemental and isotopic data suggest that the chemical variations of the LT and HT lavas cannot be explained by crystallization from a common parental magma. Instead, they may originate from different mantle sources under various melting conditions and underwent distinct differentiation and contamination processes. REE inversion calculations indicate that the HT magmas were generated by low degrees of partial melting (1.5%) of a mantle source that has εNd(t) of ∼+5 and 87Sr/86Sr(t) of ∼0.704 within the garnet stability field. These magmas were then subjected to shallow level gabbroic fractionation, which led to larger chemical variations. In contrast, parental magmas of the LT type were generated by higher degree of partial melting (16%) of a distinct mantle source (εNd(t)≈+2, 87Sr/86Sr(t)≈0.705) around the spinel–garnet transition zone. The chemical evolution of the LT lavas is controlled by an olivine (ol)+clinopyroxene (cpx) fractionation. The Emeishan flood basalts may result from a starting mantle plume. The petrogenesis of both the LT and HT magmas was further complicated by contamination of upper crust and lithospheric mantle. While the HT1 lavas have experienced an AFC style of contamination in the upper crust, the HT2 lavas that mark with U–Th depletions may result from additional interaction with melts derived from a gabbroic layer near the crust–mantle boundary. In contrast, a temperature-controlled style of contamination was associated with the LT lavas. Our data show that both temporal and spatial geochemical variations exist in the Emeishan flood basalt province. The occurrence of thick LT lavas in the western part of the province may record the main episode of the flood basalt emplacement. In contrast, the less abundant overlying HT basalts may imply a waning activity of the plume. In fact, the HT basalts are the dominant magma type in the periphery of the province. The lower degrees of mantle melting of the HT lavas may be a result of relatively thicker lithosphere and lower geotherm.