Evolution of the sub-cratonic mantle seen in mantle xenoliths

Abstract

Detailed investigations of the petrology and geochemistry of the Jagersfontein mantle pendotite xenolith suite have yielded new refinements to models of subcratonic mantle petrogenesis. The Jagersfontein peridotitic xenoliths show three distinct xenolith types as recognised by Winterburn (1987): a coarse low-temperature, a coarse medium-temperature, and a deformed high-temperature suite. Geothermobarometry, using the orthopyroxene-clinopyroxene solvus thermometer, and aluminium in orthopyroxene in equilibrium with garnet barometer of Brey and Kohler (1990), has been used to refine the depth ranges of these suites, and yields a geotherm lacking an inflection at high temperatures. The general major-minor-trace element geochemistry of garnets shows variations related to depth, whilst complex heterogeneities in the composition of single garnets yield evidence of changing geochemistry with time. By combining these types of information a model of geochemical evolution involving melt metasomatism, is developed in both space and time for the sub-cratonic mantle. Garnets from coarse low-temperature xenoliths are G9 (lherzolitic) and plot in a confined region near the base (low Ca-Cr) of the G9 trend on the Cr 203-CaO diagram. Garnets from coarse-medium temperature xenoliths are G 10 (harzburgitic) and plot in variable Ca and Cr to the low-Ca side of the G9 trend. Garnets from deformed xenoliths are also from lherzolite assemblages and define the G9 trend, with the garnets from the lowest pressure deformed samples plotting at the high Ca-Cr end of the G9 trend. With increasing depth deformed xenolith garnets plot at progressively lower Ca-Cr, with the deepest samples plotting at the base of the trend. Some garnets in the deformed, and coarse medium-temperature suites possess preserved major and/or trace element heterogeneities formed by metasomatism which diffusion constraints indicate to have occurred within a few million years prior to kimberlite eruption. Geometries of these heterogeneities give clear evidence that peridotite containing these garnets was infiltrated by a melt which caused growth of new garnet, as well as cracking/annealing, and chemical exchange. The metasomatic melt apparently originated from the base of the mantle sampled by the deformed xenolith suite, and may be associated with melt pools which precipitated the Cr-poor megacryst suite. This melt percolated upwards through the deformed xenolith mantle, evolving in composition through a combined process of crystal fractionation and crystal-liquid exchange. The major element composition of the fluid was buffered to 'lherzolitic' compositions by interaction with the peridotite matrix through which it flowed, with new garnet growth characterised by Ca-Cr compositions which move toward those of garnet megacrysts in the case of deformed G9 garnets, and toward the lherzolitic G9 trend in the case of coarse GlO (harzburgitic) garnets. The REE composition of melts calculated to be in equilibrium with garnet rims show a progressive increase in LREE and decrease in HREE with decreasing depth through the deformed, and into the coarse medium-temperature mantle from a starting composition in equilibrium with Cr-poor garnet megacrysts. This compositional evolution is dominantly controlled by fractional crystallisation of garnet, olivine, and pyroxene, from the melt as it percolated upwards. Peridotite/melt exchange probably was an important secondary process with the deformed xenolith mantle acting as a chromatographic column. The minor element Ti has a more complex behaviour, rising in abundance with decreasing depth, until the boundary between the deformed and coarse mantle is reached, at which point it appears to have been largely removed from the upwelling melt. The REE compositions of garnet cores show that metasomatism has actually depleted the peridotite in incompatible elements, through addition of HREE, and loss of LREE. Melts in equilibrium with garnet cores also show progressive LREE/HREE increases with decreasing depth, and present strong evidence that the Jagersfontein mantle has been subjected to at least one previous metasomatic event by a melt more evolved than that which precipitated the Cr-poor megacrysts. The metasomatic melt compositions are consistent with the 'anomalous' REE patterns commonly reported for garnets found as inclusions in diamonds and coarse xenoliths. These metasomatic melts were therefore responsible for the incompatible element enriched chemistries typical of garnets from the coarse-medium temperature suite, and garnets from modally metasomatised rocks from the coarse low-temperature suite. Garnets from xenoliths of the coarse low-temperature suite, which show no evidence of modal metasomatism, are depleted in incompatible elements compared to all other samples, and may represent compositions resulting from melt extraction during the Archaean. Metasomatic infiltration such as that documented by the zoned garnets, may have destroyed the harzburgitic craton root (represented by the coarse mediumtemperature xenoliths) over time. The top of the deformed mantle, characterised by lherzolitic garnets of high-Cr composition, may represent original harzburgitic material from the craton root which has been buffered to lherzolitic compositions over time by interaction with melts. This type of process is demonstrated by the zoning present in some G 1 garnets which approach G9 compositions on rims. The deformed lherzolite mantle below the harzburgitic craton root provided a column through which melt chemistries evolved. This deformed lherzolite may represent the thermal boundary layer between lithosphere and asthenosphere, being largely asthenospheric in geochemical character, but not forming part of the convecting asthenosphere beneath.