Trace elements in gem diamond from Akwatia, Ghana and DeBeers Pool, South Africa

Monocrystalline gem-quality diamonds from Akwatia, Ghana and De Beers Pool, South Africa were analyzed for their trace element content using an ultra-sensitive ‘off-line’ LA-ICP-MS method. Concentrations of 27 elements analyzed range from ppt to ppm level but mostly are below the LOQ. CeN/EuN (1–6; N = chondrite normalized) and CeN/TiN (0.6 to 12) indicate mildly elevated LREEN/MREEN and variable LREEN/TiN. One diamond from Ghana is distinct (G103) by having a much steeper LREEN/MREEN slope than the other samples and overall matches the trace element pattern of low-Mg carbonatitic melt inclusions in fibrous diamonds from Kankan, Guinea (Weiss et al., 2009). Syngenetic garnet inclusions indicate that the diamond growth medium must have been highly enriched in LREE, with CeN/EuN and CeN/TiN from 9 to 370 and 10 to 3400, respectively. This extreme discrepancy in trace element ratios between these gem diamonds and inclusion-based melt compositions is inconsistent with the commonly made interpretation that trace impurities in gem diamond represent trapped inclusions of the diamond forming fluid/melt.Mixtures of submicroscopic inclusions of common peridotitic and metasomatic phases in equilibrium with a basaltic melt mimic the relative Ce, Eu and Ti abundances of the studied diamonds. But models are dominated by orthopyroxene and hence, do not correspond to modal relationships expected for peridotitic sources. The models also require at least 0.5% of the sample volume be occupied by inclusions, which would affect diamond transparency. Incorporation of exotic minerals that are normally rare or as yet undiscovered in peridotite (e.g. crichtonite group minerals and REE phosphates) could significantly lower the required inclusion abundance to an extent that preserves the transparent nature of gem diamonds. The principal finding of our study is that, with one exception, the trace element patterns of these gem diamonds do not yield a faithful representation of the diamond growth medium and hence caution is required when extrapolating gem quality diamond trace element measurements into “fluid” compositions.

► Off-line ICP-MS trace element measurements of Ghana and S. Africa gem diamonds. ► LREEN/MREEN in diamonds lower than garnet inclusion-based fluid/melt models. ► Garnet peridotite micro-inclusion models inconsistent with diamond transparency. ► Trace mantle phase inclusions are possible element impurity hosts.