U-Pb and Lu-Hf isotopes in baddeleyite and zircon megacrysts from the Mbuji-Mayi kimberlite: constraints on the subcontinental mantle

Megacrysts of baddeleyite (ZrO2) and zircon (ZrSiO4) from the diamond-bearing Mbuji-Mayi kimberlite were analyzed to determine their age, origin, and mantle source characteristics. Two of the five baddeleyites studied show 90 degrees twins on the 0.05-0.2 mm scale suggesting formation at pressures greater than or equal to 4.5 GPa, while zircon reveals a mosaic-like structure, indicative of a sudden pressure release. The 45 U-Pb analyses define an array that intercepts the concordia curve at 69.8 +/- 0.5 (2 sigma) and 2528 +/- 452 Ma. Initial epsilon-Hf values for an age of 70 Ma are +8.4 and +8.1 for zircon and +5.1, +6.0, +6.2, +6.5, +10.2 for baddeleyite. Average Zr/Hf ratios in both minerals are almost a factor two higher than those of primitive mantle, MORB or continental crust. Baddeleyites have uranium concentrations that are exceptionally high for mantle-derived grains (780-2050 ppm). Taken together, the data indicate that zircon and baddeleyite crystallized coevally at 70 Ma in mantle reservoirs that experienced different time-integrated LILE-fractionation histories, Some limited secondary enrichment of strongly depleted mantle may have contributed to the distinct range of Hf signatures. Small amounts (<5%) of 2.5 Ga old radiogenic Pb detected in both minerals are probably inherited from pre-existing crystals in the mantle, and high Zr/Hf ratios most likely reflect very small degrees of Iherzolite melting, possibly in association with mantle metasomatism by carbonate-rich fluids. Formation of the zircon and baddeleyite megacrysts can be explained either by pre-kimberlite crystallization from different magmas or subsolidus reaction during subduction of differentiated material such as oceanic crust (+sediments?). This latter interpretation would be in agreement with the fact that inclusions in Mbuji-Mayi diamonds, and nodules in the kimberlite are dominantly eclogitic in nature. To produce the kimberlite, and to concentrate the megacrysts in a single pipe, subsequent melting of intermingled mantle domains seems the most plausible mechanism. Moreover, formation and residence times of the evolving kimberlite magmas must have been long enough to allow extraction, re-crystallization and beginning resorption of zircon and baddeleyite. (C) 1997 Elsevier Science B.V.