Miocene Volcanism of the Baikal Rift across the Boundary of the Siberian Craton: Evidence for Lithospheric Mantle Melting

Continental rifting is usually viewed in terms of two contrasting models of active and passive extension. The origin of the Baikal Rift, adjacent to the southern part of the Siberian Craton, has been described by both models in the past. It is expected that basaltic magmatism in an active model scenario should be primarily sourced from a mantle plume or plume-fed asthenosphere, whereas melting of the lithospheric mantle is expected to be a predominant source for magmatism in the passive model. In this paper, we focus on the Miocene volcanic rocks sampled along two 60 km-long profiles that cross the boundary between the Neoproterozoic Tuva-Mongolian massif and the Archean-Paleoproterozoic Siberian Craton. Most of the samples studied are trachybasalts. In terms of trace element concentrations normalized to primitive mantle, the lavas mimic OIB-like patterns with troughs at Rb, Th–U, Pb, and Y, and peaks at Ba, Nb, Ta, K, and Sr. Moreover, similar trace-element patterns to the studied samples are also observed for Miocene and Quaternary lavas located in the southwestern of the Baikal Rift, and adjacent regions of non-rifted Mongolia. According to the ratio of CaO to MgO, and TiO2/Al2O3 to SiO2, the compositions of the studied lavas coincide with experimental melts derived from mafic lithologies. Trace element data of samples suggest that garnet was a residual phase during partial melting. The Sr-Nd isotopic characteristics of the studied lavas are 87Sr/86Sr 0.70427–0.70469 and 143Nd/144Nd 0.51267–0.51284. They are identical to the coeval Miocene lavas of neighbouring volcanic fields, but they differ from the Quaternary lavas that extend to lower 87Sr/86Sr (0.7038–0.7044) with near identical 143Nd/144Nd. Isotopes of Hf for studied samples show values εHf = 6.0–7.7, except for the two samples taken within the boundary between two lithospheric blocks with εHf 4.6 and 4.8. The δ18O of olivine from lava samples is everywhere higher than that of the asthenospheric mantle and ranges from 5.5 to 6.4‰. Variations of δ18O versus Mg#, 87Sr/86Sr and εHf in the studied samples do not correlate, but do unequivocally rule out crustal assimilation. The isotopic variations are consistent with recycling of mafic crustal lithologies at mantle depths. Lavas from the Tuva-Mongolian massif and the Siberian Craton differ in lead isotopes by lower values of 206Pb/204Pb (< 17.785) and higher values of Δ8/4Pb (61–75) for on-cratonic samples and the reverse relationship for off-cratonic lava (> 17.785 and 55–61) respectively. The equation for Δ8/4Pb=[208Pb/204Pb-(1.209*(206Pb/204Pb) +15.627)] *100 is from (Hart, 1984). The correlation of lead isotopes with the mafic recycled component, the sharp change of lead isotopic values at the cratonic boundary and decoupling of lead isotope ratios from other isotopic ratios lead us to suggest that the values of 206Pb/204Pb and Δ8/4Pb are associated with an ancient accessory mineral phase such as sulphide confined within the lithospheric mantle. The predominant role of the lithospheric sources in the formation of the Miocene volcanic rocks, indicate that the volcanism of the Baikal Rift was caused by a passive tectonic process, rather than active rifting.