Melting of the Continental Crust: Products and Processes

Earth's continental crust is unique compared to the 'basaltic' crusts of other inner planets; this is a consequence of free water on Earth and recycling of Earth's primary basaltic crust. Although there is no agreement about the average chemical composition of bulk continental crust or about secular change in composition, there is agreement that upper crust is more felsic than lower crust. Further, it is widely accepted that intracrustal differentiation by melting is responsible for the difference in composition, even though controversy remains about whether part of the inventory of lower crust has been lost by delamination. Magmatic additions from the mantle, recycling through erosion, sedimentation and burial via subduction and/or orogenesis, and regeneration through multiple orogenic events have developed a diversity of crustal rocks that neither melt nor behave uniformly. Water is an important constituent of crustal rocks, but the lower crust may be essentially 'dry', so that melting is controlled by successive dehydration of muscovite, biotite and hornblende in bulk compositions that vary from 'granodioritic' or 'pelitic' to 'dioritic' or 'basaltic'. Thus, the products of melting of continental crust are varied, both in terms of melt compositions, and in terms of the residue. Further, since H2O is partitioned in a hydrous silicate melt, cooling of melt bearing crust in a closed system potentially will lead to retrogression by back reaction between melt and residue, unless segregation has separated them sufficiently to avoid reaction during cooling. This raises the issue of equilibration domains and local equilibrium. The common occurrence of pristine to weakly retrogressed residual granulites suggests melt loss from the lower crust, consistent with the presence of leucogranites in the upper crust. We illustrate the 'reaction principle' (Bowen, N.L., 1922, The Reaction Principle in Petrogenesis. Journal of Geology, 30, 177-198) with reference to microstructures in natural systems, and by the use of pseudosections [including T-X and P-X pseudo-binaries (or projections) of fixed bulk compositions] modeled in the MnNCKFMASH system under suprasolidus conditions using an internally consistent thermodynamic database and the software THERMOCALC. Such diagrams are potentially powerful interpretative tools, and it is instructive to consider the consequences of a range of possible paths through them. For example, univariant reactions are rarely intersected ('seen') by single rocks (fixed bulk compositions) during burial and exhumation, even in relatively simple sub-systems (e.g., KFMASH). Modeling melt loss from metapelitic compositions in the MnNCKFMASH system shows that both bulk composition and proportion of melt lost give rise to a wide variation in retrograde behavior consistent with observation of lower crustal rocks. This point is important in understanding evolution of the melts produced, since retrograde (back) reactions occur at peritectics in the system. Thus, modeling of fractionation processes must take into account the possibility of such reactions and the consequent change in phase asssemblages. In the absence of a robust thermodynamic model for sapphirine, and for Ti in model granite melt compositions, we are not yet able to model metapelite compositions under ultra-high temperature metamorphic conditions. Further, in the absence of a robust thermodynamic model for the amphiboles, we are not yet able to model crustal melting of metabasaltic compositions. Experimental data are available that partly remedy this deficiency and provide some information about melt compositions and solid residues, but only for a limited range of fixed bulk compositions or for compositionally restricted sub-systems.