Geochemistry of Intermediate/- to High-Pressure Granulites

Abstract

Geochemical data from the literature for high-pressure granulites have been compiled with the aim of evaluating the compositional differences between granulite terrains of various ages and xenoliths and the processes responsible for granulite formation. Although complete compositional overlap exists between the different granulite groups, median compositions of Archean granulite facies terrains are more evolved (having higher SiO2 and lower MgO) than post-Archean terrains, which are in turn more evolved than granulite xenoliths. The degree of LREE enrichment changes systematically as well, with Archean terrains having the highest (La/Sm)N and (La/Yb)N and xenoliths the lowest. In contrast to the secular changes observed in upper crustal composition, the median K2O content for Archean granulites is slightly higher than that of post-Archean terrains. K, Rb, Cs, Th and U distribution patterns of granulites are the same for terrains and xenoliths, suggesting similar depletion processes operate in both. It is possible to classify granulites on the basis of their LILE contents and thereby to predict their LILE characteristics if their K contents are known. K/U and K/Th ratios are significantly higher in granulites than in the upper continental crust and do not correlate with K content. Using the median values of K, K/Th and K/U for Archean granulite terrains, a heat production of 0.48 µW/m3 is calculated — a value similar to estimates based on individual terrains. Suggestions that mantle heat flow is overestimated in Archean shields due to the effects of lateral heterogeneities in heat-producing elements in the crust mean that Archean granulite terrains cannot be excluded as being representative of the lower crust in these regions on the basis of heat flow arguments. The heat production calculated for post-Archean terrains is somewhat higher (0.53 µW/m3) than the Archean value and that of xenoliths is very low (0.08 µW/m3). Model trace element contents of partial melting residues of metapelitic rocks, using recent experimental results to constrain restite phase proportions, are different from those observed in aluminous granulites. However, if melt segregation was inefficient, the trace element characteristics of the model restite are closer to those observed in aluminous granulites, but the residues are no longer LILE depleted. This suggests that (1) granite residues may be sampled as undepleted granulites, and (2) partial melting is not the process by which granulites become depleted in LILEs. Finally, positive Eu anomalies postulated to exist in the lower crust are observed mainly in cumulates, suggesting that crystal accumulation rather than partial melt removal may be the process responsible for imparting the negative Eu anomaly on the upper crust.