| Monocrystalline gem-quality diamonds from Akwatia, Ghana and De Beers Pool, South Africa have trace-element concentrations ranging from ppt to ppm levels, but mostly are below the limit of quantification. CeN/EuN (1--6; N = chondrite normalized) and Ce N/TiN (0.6 to 12) indicate mildly elevated LREE/MREE and variable LREE/Ti. Syngenetic garnet inclusions indicate that the diamond growth medium must have been highly enriched in LREE, with CeN/Eu N and CeN/TiN from 9 to 370 and 10 to 3400, respectively. One sample, G103, has trace-element characteristics that closely resemble a carbonatitic fluid. The remaining samples show discrepancies in trace-element ratios between these diamond and inclusion-based fluid/melt compositions is inconsistent with the generalized interpretation that trace impurities in gem diamond represent trapped inclusions of the diamond-forming fluid/melt.;Gem-quality micro-diamonds from the Panda kimberlite (Ekati, Canada) have delta13C values that are on average 1.3‰ higher than macro-diamonds from the same pipe. This documents either distinct diamond-forming fluids, fractionation process, or growth histories for these two populations. A broad trend to more 13C enriched compositions with decreasing mantle residence temperature (proxy for decreasing depth) is interpreted to reflect diamond formation from a carbonate-bearing metasomatic fluid/melt that isotopically evolves as it percolates upward through the compositionally layered Central Slave cratonic lithosphere.;The linear relationship between platelet peak area and NB concentration that defines "regular", non-cuboid diamonds is evaluated with a world-wide database of FTIR diamond data. This relationship is expressed as I(B') = 0.61NB and can be projected with confidence to at least ~550 at. ppm NB. This database also shows that maximum hydrogen-related IR absorbance (3107 cm-1 center) correlates with increasing NB concentrations, implying a relationship between IR-active hydrogen at 3107 cm-1 and the aggregation process that forms B-centers in diamond.;Model mixtures of submicroscopic inclusions of common peridotitic and metasomatic phases fail to mimic expected modal relationships for peridotitic sources. Additionally, the calculated inclusion abundances in these models would affect sample transparency, which is not observed. Trace-element patterns for most of the studied gem diamonds are not a direct representation of the diamond growth medium. |
