The origin, composition, and evolution of the Kimberley kimberlites (South Africa)

Abstract

Kimberlites are deeply derived (i.e., >150 km), small-volume igneous bodies that have been emplaced on all continents throughout the last 2.8 billion years. The typical volcanic expression of kimberlites is a deep irregular root zone and/or feeder dyke system connected to a regular steeply dipping and outwardly tapering pipe-like diatreme, which may be overlain by a crater and extrusive material (when not removed by erosion). The crater and diatreme facies contain pyroclastic rocks, which transition into coherent (sub-volcanic) rocks in the root zone. Kimberlites are of economic value as the major host of gem quality diamonds at the Earth’s surface. They also hold great scientific significance, as the deepest derived melts to reach the surface, with entrained mantle material that provides some of our best information on the structure, composition, and evolution of the sub-continental lithospheric mantle. However, despite over a century of dedicated research, numerous aspects of kimberlite petrogenesis remain poorly understood and contentious. One central issue that this project has addressed is the composition and evolution of kimberlite melts. The composition of kimberlite melts remain poorly constrained because: (1) rocks emplaced near the surface are prone to deuteric and hydrothermal alteration; (2) they have been contaminated by the physical incorporation of xenocrystic and xenolithic material; (3) their parental magmas have been modified by interaction with and partial assimilation of mantle and crustal material; and (4) they undergo syn-emplacement differentiation. Therefore, in this study exceptionaly ‘fresh’ rocks from the well studied kimbeley cluster (the type locality) were examined to gain further insights into kimberlite melt compositions. Better constraints on kimberlite melt compositions are pivotal if we are to move toward a comprehensive understanding of the petrogenesis of these enigmatic rocks.

The studied samples derive from the Kimberley cluster (South Africa), which lies within the Western terrane of the Kaapvaal carton. This cluster constitutes the type locality of kimberlites, containing five major kimberlite pipes (The Kimberley mine, De Beers, Dutoitspan, Wesselton, and Bultfontein), numerous smaller pipes, and abundant dyke/sill complexes (e.g., Benfontein, Wesselton Floors, Wesselton Water Tunnels). The Kimberley cluster has been dated by various geochronological techniques yielding emplacement ages of ~80-90 Ma. To provide new insights into the composition and evolution of kimberlite melts, a detailed petrographic study of sub-volcanic (hypabyssal) coherent kimberlites was conducted. This included the investigation of mineralogy, mineral zonation, inclusion populations (mineral, melt, and fluid), and textural relationships between phases, utilizing a range of microscopy techniques. This petrographic data formed the basis of targeted geochemical analysis by electron microprobe. A study of olivine compositions across multiple intrusions of the Kimberley cluster shows that olivine, more than any other mineral, provides the most complete record of kimberlite evolution. This study showed that pre-ascent metasomatism of the lithosphere by kimberlite melts is wide-spread, and that so-called ‘xenocrystic’ olivine is not directly representative of the wider lithosphere due to metasomatism of the conduit by previous pulses of kimberlite magmatism. The composition of kimberlitic liquidus olivine overlaps that of olivine from other mantle-derived carbonate-bearing magmas (orangeites, ultramafic lamprophyres, melilitites), with low Mn/Fe and Ca/Fe, and moderate Ni/Mg ratios. It is suggested that these compositions are typical of olivine in equilibrium with melts derived from carbonate-rich peridotite sources. Compositional zonation patterns indicte that olivine crystallises throughout magma ascent and that its crystallisation continues after emplacement into the upper crust. Magmatic olivine (i.e. crystallised from the kimbelrite magma) displays distinct generations of crystallisation, with increasing Mg, Ca, and Mn contents interpreted as the result of fractional crystallisation and increasing oxygen fugacity (fO2). The stability of olivine at sub-solidus conditions implies that secondary melt inclusions cannot trap primitive melts, but rather evolved residual fluids. Although olivine provides a wealth of information on early kimberlite melt evolution and metasomatism of the surrounding lithosphere, details about the later stages of kimberlite melt evolution are evident in other magmatic groundmass phases. Therefore, detailed petrographic and mineral chemical studies were undertaken on all mineral consistents from a suite of samples from the exceptionally fresh De Beers dyke to determine the crystallisation sequence. In turn, this data yielded insights into melt evolution from the perspective of multiple different magmatic phases. The early stages of kimberlite crystallisation (i.e., olivine, Cr-spinel, Mg-Ilmenite, rutile) are defined by decreasing Mg/Fe ratios. This subsequently reverses (i.e., increasing Mg/Fe) during later groundmass crystallisation, which is attributed to increasing fO2. Comparison with published data shows that the melt parental to early crystallising phases in this dyke are indistinguishable from those in the root-zone intrusions of the Kimberley cluster, meaning that not all dykes are the crystalisation product of magmas that underwent pre-emplacement fractionation. To gain additional insights into the very late stages of kimberlite melt evolution further detailed petrographic and mineral chemical studies were conducted on late-stage groundmass phases (i.e., apatite and mica) from samples of different root zone intrusions and dyke/sill complexes in the Kimberley area. Despite the early crystallising phases (i.e., olivine, Cr-spinel, Mg-ilmenite) being compositional indistinguishable in dykes/sills and root zone kimberlites, the compositions of apatite appear to be controlled by the style of magma emplacement. Apatite from dykes/sills is Si-rich and Sr-poor, whereas apatite in root zone intrusions show the opposite features. The high Si content of apatite in dykes/sills is attributed to the coupled incorporation of silica and a carbonate ion for phosphorus, reflecting higher CO2 contents in the melts parental to dykes/sills. The high Sr content of apatite in root zone intrusions likely requires crystallisation from, or overprinting by, hydrous fluids. These features indicate that dyke/sill kimberlites have higher CO2/H2O ratios than the magma that produced root zone intrusions. This is consistent with petrographic observations, whereby dykes/sills are enriched in carbonates, may contain dolomite, and have lower abundances of serpentine, mica, and monticellite than root-zone kimberlites. These differences in CO2/H2O ratios of the crystallised melt are attributed to differences in emplacement style, whereby a rapid decrease in pressure in root zone kimberlites leads to exsolution of a (CO2-rich) fluid phase, possibly caused by breakthrough to the surface. The knowledge gained through detailed petrographic and mineral chemical studies led to the development a new quantitative method for reconstructing the composition of kimberlite melts. This model allowed for constraints to be placed on the composition of primitive kimberlite melts and their evolution, as they incorporate and assimilate xenocrystic material, undergo fractional crystallisation, and post emplacement alteration. The results of this modelling indicate that the melt parental to the Bultfontein kimberlite was transitional between silicate and carbonate. This composition is consistent with experimental constraints on the amount of CO2 that can be dissolved into kimberlite melts in the upper crust. The reconstructed primitive melt composition is in equilibrium with asthenospheric source rocks. Based on constraints from experimental studies, Kimberley kimberlites could have been produced by ~0.5% melting of carbonated lherzolite in the upper asthenosphere (i.e., 6.0-8.6 GPa and ~1400-1500 degrees Celsius).