Steven B. Shirey

Rhenium-osmium (Re-Os) Studies of Sulfide Inclusions in Diamonds from the Orapa Kimberlite, Kaapvaal Craton, Botswana

Sulfides, chiefly of iron (Fe) but also with subordinate nickel (Ni) and copper (Cu), are common inclusion minerals in diamond that occur in two basic assemblages: peridotitic (p-type) and eclogitic (e-type). At DTM the recent development, by former fellow Graham Pearson, Richard Carlson, and me, of microchemical techniques for the analysis of the rhenium-osmium isotopic compositions of single sulfide inclusions at 10^-15 gram levels has led to detailed geochronology on individual diamonds that contain either inclusion assemblage. Apart from the obvious economic importance of such work to diamond exploration, any materials encased in diamonds provide a unique geological sample of mantle minerals that have been isolated from chemical exchange since diamond growth. Diamond distribution on Earth is closely linked to the distribution of continental mantle keels. Thus, an important focus of this research is to compare diamond growth episodes to geological processes such as continental core (craton) stabilization and much later continent-margin accretion via subduction to understand the role of continental keels in continent development.

A Re-Os isotopic study of sulfide inclusions in diamonds from the 93-million-year-old Orapa kimberlite in Botswana is under way. The Orapa kimberlite is host to one of the world's largest and most productive diamond mines and is situated on the northwest side of the ancient continental core making up southern Africa called the Kaapvaal craton. Orapa is noted for its well-exposed kimberlite (the diamond host rock), its diamond-bearing and non-diamond-bearing eclogite (high-pressure metamorphosed basalt) xenoliths (rock fragments carried by the kimberlite), and the high proportion of sulfide and silicate in its diamond inclusion population. In general, we want to better understand the age spectrum of diamonds containing e-type sulfide inclusions at localities such as Orapa, where we can compare sulfide inclusion ages to previously studied eclogitic silicate inclusion ages. A goal is to look for other generations of diamond growth, and to explore the relationship between the sulfide inclusions in diamond and their eclogite xenolith hosts in ways that can be extended to the larger Kaapvaal craton and its geological history.

The sulfides fall into two groups based on the Re-Os isotopic data: one group with a 2,900-million-year age and the other with a 990-million-year age (Fig. 12). The older age represents the first firm Archean age for diamonds containing eclogitic materials. This is exciting because it begins to resolve the long-standing issue of why there have been good examples of old diamond-bearing eclogite xenoliths but few examples of old eclogitic diamond inclusions of either sulfide or silicate type. The younger age is interesting because it is identical to the previous 990 million-year-old samarium-neodymium isochron age obtained on Orapa eclogitic garnet and clinopyroxene diamond inclusions.

These data suggest that there were at least two episodes of diamond growth in the mantle beneath the Orapa area, separated by about 2,000 million years. The striking similarity of the older ages to the rhenium-depletion model ages on peridotites from the Letlhakane kimberlite 40 km to the east-southeast of the Orapa kimberlite makes a compelling argument that continental core stabilization about 2,900 million years ago was accompanied by oceanic crustal underthrusting to make eclogite. The striking similarity of the younger ages to the eclogitic sulfide inclusions in diamond at the Koffiefontein kimberlite some 800 km to the south indicates that eclogite emplacement into the cratonic lithosphere 1,000 million years ago may be more widespread than previously thought.

Fig. 12. This Re-Os isochron diagram shows Orapa and Koffiefontein eclogitic (e-type) sulfide inclusions in diamond. Each data point represents the inclusion from a single diamond. Error bars are derived chiefly from blank correction uncertainties when analytical concentrations of Os are low (e.g., 30-100 x 10^-15 grams). Error bars are not shown when smaller than symbols. 990 and 2,900 million-year-old (Ma) lines are shown for reference purposes. (Koffiefontein data from Graham Pearson of the University of Durham and coworkers.)