Generating early continental crust
Archean cratons are the products of geodynamic processes governing the evolution of the early Earth. They comprise felsic granitoids enclosed by folded volcano-sedimentary successions. However, these rocks may not be representative archives but rather accidental fragments in time and space. The further back the rock record reaches in time, the more ambiguous it becomes, not least because of multiple subsequent overprinting events. The composition of Archean granitoids is distinctly different to post-Archean felsic rocks that are associated with modern continental crust forming processes. Most common is an association of tonalites trondhjemites and granodiorites (TTG). The joint occurrence of TTG potentially holds key insights about the Archean geodynamic environment, including the timing of the onset of modern-day plate tectonics, which is yet to be resolved. A major uncertainty regarding their formation is the availability of H2O during melting of their protolith and the mechanism of their differentiation.
This research focusses on constraining P-T-H2O conditions of TTG melt formation by combining different approaches: (i) an experimental Ti saturation study on an Eoarchean granodiorite from the Nuvvuagittuq greenstone belt, North-eastern Superior Province and subsequent calibration of a Ti solubility model for silicate melts, (ii) an experimental and modelling investigation of saturated liquidus phases and potential cotectic paths on the same rock and (iii) using published glass analyses of hydrous melting experiments to calibrate a support vector machine regression (SVR) model capable to predict temperature, pressure and H2O conditions of a TTG sample based on the major element composition. The three methods are applied to a compilation of natural TTG analyses from published literature. Ti saturated liquidus temperatures for natural TTGs are on average between 750 and 900 ◦C, constraining a minimum melt formation temperature. However, many TTGs have temperatures below expected liquidus temperatures based on their major element composition. Congruously these can not have been saturated with rutile or ilmenite at their liquidi. This has consequences for the interpretation of the Nb depletion in TTGs, as it is observed, irrespective of whether or not a composition could have been saturated in Ti. Therefore the role of Ti accessory phases in TTG formation might be limited. The Nb depletion could instead be inherited from a precursor rock or the result of other Ti-bearing phases involved in TTG formation, as for example amphibole or mica.
Based on comparison with experimental data, TTGs in equilibrium with a Ti phase at the liquidus require 8-12 wt% H2O dissolved to be at liquidus conditions. This elevated amount favours the presence of a free fluid during melting. The SVR model independently confirms the amount of H2O required for Ti saturated liquidus temperatures.
Hydrous conditions are also favoured when retracing the differentiation paths of TTG via crystallisation experiments and modelling. Plagioclase suppression by H2O contents >5 wt% results in a low K2O/Na2O ratio throughout differentiation that is typical for melt evolution towards a trondhjemitic composition. Less hydrous conditions result in granodioritic and granitic melt compositions, as is commonly observed in post-Archean magmatic systems. Therefore the Archean environment producing TTGs must have been more hydrous than recently active magmatic systems that produce felsic crust. It is most likely that TTGs do not align along a single cotectic path, but along several hbl/cpx - grt/opx controlled cotectics, whose position is shifted by varying pressure and/or H2O conditions.
The geodynamic environment that gave rise to TTGs must provide an elevated amount of H2O. This favours a model of the recycling of hydrated basalt into the region of melt formation in a subduction-like setting.