Partial melting and the formation of granulite facies assemblages in Namaqualand, South Africa

Abstract Dehydration-melting reactions, in which water from a hydrous phase enters the melt, leaving an anhydrous solid assemblage, are the dominant mechanism of partial melting of high-grade rocks in the absence of externally derived vapour. Equilibria involving melt and solid phases are effective buffers of aH₂,o. The element-partitioning observed in natural rocks suggests that dehydration melting occurs over a temperature interval during which, for most cases, aH₂o is driven to lower values. The mass balance of dehydration melting in typical biotite gneiss and metapelite shows that the proportion of melt in the product assemblage at T± 850°C is relatively small (10–20%), and probably insufficient to mobilize a partially melted rock body. Granulite facies metapelite, biotite gneiss and metabasic gneiss in Namaqualand contain coarse-grained, discordant, unfoliated, anhydrous segregations, surrounded by a finer grained, foliated matrix that commonly includes hydrous minerals. The segregations have modes consistent with the hypothesis that they are the solid and liquid products of the dehydration-melting reactions: Bt + Sil + Qtz + PI = Grt ° Crd + Kfs + L (metapelite), Bt + Qtz + Pl = Opx + Kfs + L (biotite gneiss), and Hbl + Qtz = Opx + Cpx + Pl + L (metabasic gneiss). The size, shape, distribution and modes of segregations suggest only limited migration and extraction of melt. Growth of anhydrous poikiloblasts in matrix regions, development of anhydrous haloes around segregations and formation of dehydrated margins on metabasic layers enclosed in migmatitic metapelites all imply local gradients in water activity. Also, they suggest that individual segregations and bodies of partially melted rock acted as sinks for soluble volatiles. The preservation of anhydrous assemblages and the restricted distribution of late hydrous minerals suggest that retrograde reaction between hydrous melt and solids did not occur and that H₂O in the melt was released as vapour on crystallization. This model, combined with the natural observations, suggests that it is possible to form granulite facies assemblages without participation of external fluid and without major extraction of silicate melt.