Generation of felsic crust in the Archean: A geodynamic modeling perspective
Graphical abstract
Introduction
The greater rate of production of continental crust in the Archean (e.g. Dhuime et al., 2012), and the occurrence of tonalite, trondhjemite and granodiorite complexes, and komatiites, which are largely restricted to the Archean (Goodwin, 1991), are consistent with a hotter Earth. Based on the most primitive liquidus temperatures (Abbott et al., 1994) or mantle potential temperatures (Herzberg et al., 2010) derived by inversion of the chemistry of non-arc basalts from greenstone belts and calculations of the thermal evolution of Earth (Korenaga, 2008a, Korenaga, 2008b, Labrosse and Jaupart, 2007), the upper mantle temperature is estimated to have been up to ∼1600 °C in the early Archean. This is ∼250 °C higher than the average at the present-day, although the present value varies from 1280 °C to 1400 °C (Herzberg et al., 2007). Higher mantle temperatures together with higher radiogenic heat production, which might have been up to 3 times higher in the Archean (e.g. Brown, 2007, Davies, 1992), will have impacted both the thickness and composition of the crust. As a consequence of secular cooling, there is generally no modern analog to assist in understanding the tectonic style that may characterize the Archean, particularly prior to 3 Ga. For this reason, numerical modeling that is constrained by the fragmentary evidence preserved in the geological record is an appropriate tool to evaluate hypotheses of Archean crustal formation.
One of the main lithological associations of Archean gray gneiss complexes is the sodic tonalite–trondhjemite–granodiorite (TTG) suite (Jahn et al., 1981). Since the first description, TTGs have been the subject of much discussion related to their petrogenesis and possible Archean tectonic regimes. Nevertheless, after forty years of investigation there are still many open questions about the generation of TTGs (see overview in Moyen and Martin, 2012). TTGs are a diverse group of silica-rich rocks (SiO2 ≈ 64 wt.%) with high Na2O contents (3.0 wt.% < Na2O < 7.0 wt.%) and correlated low K2O/Na2O (0.3–0.6). They contrast sharply with Archean potassic granitoids (Moyen, 2011) and modern granitoids, which are commonly richer in K2O (granodiorites to granites). One of the crucial points about Archean geology is that the oldest preserved felsic plutonic rocks are mostly TTGs, particularly before 3.2 Ga, while potassic granitoids appear later in Earth history, locally after 3.2 Ga, e.g. in the Barberton area (Kamo and Davis, 1994), and globally by the end of the Archean (Keller and Schoene, 2012). In addition, high-Mg monzodiorites and granodiorites (sanukitoids), which are thought to derive primarily by hybridization between mantle peridotite and a component rich in incompatible elements, occur in many late Archean terrains (Martin et al., 2005). Based on these secular changes in the typology of granitoids, Laurent et al. (2014) proposed a transition to a global plate tectonics geodynamic regime during the late-Archean (3.0–2.5 Ga). This last proposition raises one of the main unanswered questions in understanding the Earth during the Precambrian: what were the tectono-magmatic processes that prevailed in the Archean Eon, particularly during early Archean time, that were responsible for the construction of the unique Archean crust?
Although the TTGs as a group are diverse with a continuum of compositions, Moyen (2011; see also Moyen and Martin, 2012) has classified them into three types related to the depth of melting in the source: (1) a high pressure group comprising about 20% of TTGs, formed at P > 1.6–1.8 GPa (rutile present) and characterized by higher values of Al2O3, Na2O and Sr, and lower values of the HREEs (heavy rare earth elements), Nb and Ta; (2) a low pressure group comprising about 20% of TTGs, formed at P < 1.0–1.2 GPa (garnet absent) and characterized by lower values of Al2O3, Na2O and Sr, and higher values of the HREEs, Nb and Ta; and (3) a medium pressure group comprising about 60% of TTGs, with intermediate geochemical characteristics. According to Moyen and Martin (2012), only high-Al2O3 sodic granitoids with low HREEs should be named TTGs. Nevertheless, the term TTG is commonly used for a wide range of sodic plutonic rocks, in some cases even including the associated potassic granitoids.
The main geochemical features of the three types of TTGs relate to the stability of plagioclase, garnet and rutile during melting. The modal proportion of garnet in the residual assemblage in equilibrium with the melts progressively increases from <5% at 1.0 GPa, which results in less pronounced depletion in the HREEs in the low pressure group, up to ∼40% at 2.5 GPa, which leads to the pronounced depletion in the HREEs in the high pressure group (Moyen and Martin, 2012, Zhang et al., 2013). The temperature of formation of the low and medium pressure TTG melts varies from 700 °C to 1000 °C, while for high pressure TTG melts the range is from 1000 °C to 1100 °C (Moyen, 2011). Taking into account all of the characteristics of TTGs, there is a growing consensus supporting the generation of TTGs by melting of hydrous metabasalt at garnet amphibolite, granulite or eclogite facies conditions (e.g. Barker and Arth, 1976, Condie, 1986, Foley et al., 2002, Jahn et al., 1981, Martin, 1986, Moyen and Stevens, 2006, Rapp et al., 1991, Springer and Seck, 1997), although alternative models persist as discussed below.
The diversity in the composition of TTGs has led to different tectonic settings being proposed for the formation of the parental melts. Formation within a subduction zone takes into account the necessity to melt hydrated basalts at garnet amphibolite, granulite or eclogite facies conditions and allows for the interaction of TTG melts with mantle peridotites in the mantle wedge (e.g. Arth and Hanson, 1975, Condie, 1981, Hastie et al., 2015, Martin, 1986, Martin and Moyen, 2002). In this model, the high proportion of TTGs in the Archean crust is commonly explained by more extensive slab melting due to higher temperature in the subduction zone (e.g. Moyen and Stevens, 2006). Moyen and Stevens (2004) showed that the low-pressure TTGs appeared earlier in the geological record (around 3.55 Ga), while the high-pressure TTGs occurred somewhat later (3.45–3.22 Ga), which they interpreted as a change from the formation of TTGs in an intra-oceanic continental nucleus to the generation of TTG melts in subduction zones. On the other hand, similar geochemical characteristics to those associated with subduction might be expected from formation of the TTG melts by delamination of the lower crust (e.g. Bédard, 2006). During delamination, sinking blocks of mafic rock might interact with the mantle in a similar way to the interaction between the subducting plate and the overlying mantle wedge (Moyen and Martin, 2012).
Indeed, the main alternative model to subduction for the generation of TTG melts is by partial melting at depth in thickened crust or at the base of oceanic plateaux (Atherton and Petford, 1993, Bédard, 2006, Qian and Hermann, 2013, Smithies, 2000, Zhang et al., 2013). Furthermore, based on melting experiments, Qian and Hermann (2013) and Zhang et al. (2013) argue that it may not even be necessary to over-thicken the crust since the most appropriate conditions for producing low-to-intermediate pressure TTG melts from mafic lower crust are 800–950 °C at 1.0–1.25 GPa, which corresponds approximately to depths of 35–44 km (using a crustal density of 2900 kg/m3). Based on the geochemistry of Archean TTGs and the subcontinental lithospheric mantle (SCLM), Bédard (2006) proposed that TTG melts were derived from the base of thick basaltic plateaus formed above mantle upwellings (plumes in his model, but with higher mantle temperatures in the Archean these upwelling need not be deeply sourced). He argued that delamination of crustal residues after such melting could catalyze multi-stage melting of the SCLM and allow maturation of the Archean continental crust. In a subsequent development, Bédard et al. (2013) proposed a model of cratonic drift in response to mantle wind for the aggregation of Archean cratonic and oceanic terranes (basaltic plateaux), including the development of structures related to bulk regional horizontal contraction. The accretion of terranes led to thickening and delamination of mafic crust coupled with ascent of hot mantle generating voluminous pulses of coeval basalt and TTG magmas.
Lastly, Kleinhanns et al. (2003) proposed an alternative scenario for the formation of TTG melts within the suprasubduction mantle wedge by fractional crystallization under water-saturated conditions. They invoked a major role for aqueous fluid in the formation of Archean TTGs, and further suggested that as the role of aqueous fluids diminished with time, so there was a change from sodic (TTGs) to potassic granitoids during the late Archean (cf. Keller and Schoene, 2012). In a similar approach, based on extensive studies of granitoids from the Kohistan batholith, Jagoutz et al. (2013) proposed hydrous fractionation of subduction related magmas in the lower crust of arcs as model for TTG genesis in the Archean.
The remaining challenge in Precambrian geology is to develop a successful paradigm of global geodynamics and lithospheric tectonics for the early Earth, such as the plate tectonics paradigm for contemporary Earth, within which the growing number of observational and analytical data may be integrated. Consequently, the debate about geodynamic settings for formation of TTG melts, which had been centered on field relationships and geochemistry, has now been joined by a series of numerical modeling efforts that are somewhat contrasting in approach.
In an early study, Van Thienen and Van den Berg (2004) argued that a plate tectonics regime was unlikely on a significantly hotter Earth. Using a numerical thermo-chemical mantle convection model, these authors proposed that local crustal overthickening leading to the transformation of basalt to eclogite in the lower crust might trigger a resurfacing event, during which a large segment of crust over 1000 km long sank into the mantle within 2 million years. In this scenario, Van Thienen and Van den Berg (2004) identified two possible settings for partial melting of metabasalt: at the base of the new replacement crust, and in the sinking crustal material itself. These authors proposed that melting in these two settings was mostly responsible for the formation of Archean TTGs, in contrast to some other settings suggested in the literature, such as small-scale delamination of the lower crust, crustal thickening, and melting of a mantle diapir, which, they argued, do not account for the P–T conditions determined for most TTG melts. However, Van Thienen and Van den Berg (2004) admitted that ignoring emplacement of basaltic melts within or under the crust in their model might lead to underestimation of the importance of other mechanisms.
More recently, Thébaut and Rey (2013) used thermo-mechanical numerical experiments to investigate the density-driven sagduction (downwelling) of greenstones and the simultaneous rise of granite domes from the hot and weak basement beneath. They did not concentrate on the formation of the felsic basement, taking it as pre-existing. The results of their numerical experiments are consistent with the existing petrological and isotopic data from long-lived hydrothermal systems associated with steeply inclined greenstone belts, where the fluid was derived from the associated ocean. In a further development, Rey et al. (2014) have proposed that the early continents may have generated intra-lithospheric gravitational stresses large enough to drive lateral spreading of thick continental crust to initiate subduction at the continental edges. Again, the origin of the continents was not discussed.
Moore and Webb (2013) invoked a heat-pipe model for the early Earth similar to that on Jupiter's moon Io. These authors argued that as surface volcanism was fed by localized channels of rising basalt (the heat pipes) so the intervening lithosphere was advected downwards to conserve mass. Heating during sinking would lead to melting of the mafic to ultramafic crust. Thus, felsic volcanics and TTG plutons may have been sourced from the downwards-advecting lithosphere. Moore and Webb (2013) further suggest that a rapid decrease in heat-pipe volcanism might have led to initiation of plate tectonics around 3.2 Ga.
In a stagnant-lid tectonic regime the mantle driving forces do not exceed the lithospheric yield strength, resulting in a single, continuous rigid plate overlying the mantle. However, in a mobile-lid tectonic regime the mantle driving forces exceed the yield strength of the brittle lithosphere, causing it to fracture into plates that move relative to each other. Thus, a transition from one regime to the other might be expected with secular cooling. To investigate such a transition models that develop episodic subduction have started to appear in the literature during the last few years (e.g. Debaille et al., 2013, Moyen and Van Hunen, 2012, O’Neill et al., 2007a, O’Neill et al., 2007b, O’Neill and Debaille, 2014, Rey et al., 2014).
Based on paleomagnetic analysis and numerical modeling, O’Neill et al., 2007a, O’Neill et al., 2007b; see also O’Neill et al., 2015) argued for episodic plate tectonics in the Precambrian, where higher mantle temperatures would result in lower lithospheric stresses, allowing for rapid pulses of subduction interspersed with periods of relative quiescence (cf. Moresi and Solomatov, 1998). The idea of a dominantly stagnant-lid regime interspersed with short bursts of subduction is supported by some geochemical data (e.g. Debaille et al., 2013, Griffin et al., 2013). Additionally, thicker crust and lithosphere might have been a serious limitation to the initiation of subduction in the Archean. If true, a different mode of downwelling (Davies, 1992) or ‘sub-lithospheric’ subduction (Van Hunen and van den Berg, 2008) might have characterized early Earth tectonics, although the conversion of basalt to eclogite may significantly relax this limitation at some stage. Nevertheless, frequent slab breakoffs and crustal delamination may have played a more dominant role in the Precambrian (Van Hunen and van den Berg, 2008). In a complementary study, Moyen and Van Hunen (2012) argued that the late Archean geological record typically suggests short-lived subduction events, much shorter than at the present-day. They argued that magmatism with arc-like geochemistry in the Archean lasted no longer than a few tens of millions of years. Taken together, these modeling results seem to fit with geochemical observations that suggest frequent alternation of arc-style and non-arc-style volcanism on a similarly short time scale (e.g. Benn and Moyen, 2008).
For more than forty years the origin of Archean TTGs and the tectonic settings in which they could have been generated have been controversial topics. Fractional crystallization of basaltic magma was initially thought to be the dominant mechanism by which TTG magmas formed and this model retains some support in the recent literature (e.g. Arth et al., 1978, Barker and Arth, 1976, Grove et al., 2003, Hastie et al., 2015, Jagoutz et al., 2013, Kleinhanns et al., 2003). However, increasingly models based on partial melting of amphibolite/granulite/eclogite have become preferred (e.g. Barker and Arth, 1976, Condie, 1986, Foley et al., 2002, Jahn et al., 1981, Moyen, 2011, Moyen and Stevens, 2006, Qian and Hermann, 2013, Rapp et al., 1991, Rapp et al., 2003, Springer and Seck, 1997, Zhang et al., 2013). Also, whether the petrogenesis of TTG melts was associated with subduction remains a matter of debate (Bédard, 2006, Bédard et al., 2013, Foley et al., 2003, Johnson et al., 2014, Martin, 1986, Martin et al., 2014). To reduce the controversy, a new generation of models is required that addresses the genesis of the unique Archean TTG crust by incorporating geochemical, geological, petrological and geophysical data. Here we present a newly developed 2D coupled petrological–thermomechanical tectono-magmatic numerical model with parameters appropriate to early Archean conditions to address such questions. The model includes spontaneous plate generation and movement, partial melting, melt extraction and melt emplacement resulting in crustal growth, radiogenic heat production, eclogitization, and other factors. The experimental results we report here assist our understanding of the dominant geodynamic regime that prevailed in the early Archean and demonstrate a series of possible tectonic scenarios for the formation of continental silicic melts, including TTGs.
Section snippets
Description of the numerical model
For this study we used a petrological–thermomechanical numerical model based on the I2VIS code of Gerya and Yuen (2003a). The model describes a series of thermal, chemical and mechanical processes appropriate to a 2D section through the lithosphere and underlying mantle (Fig. 1). The model evolves self-consistently according to the spontaneous material redistribution in response to contrasts in densities and viscosities intrinsic to different lithologies and those induced by temperature
Experimental results
Next we present a detailed description of the reference experiment (Sections 3.1 Model evolution, 3.2 Geodynamic settings for the formation of continental crust, 3.2.1 Delamination and dripping of the lower crust into the mantle, 3.2.2 Local thickening of the crust, 3.2.3 Small-scale crustal overturns, 3.2.4 Discussion, 3.3 Long-term evolution of the continental crust, 3.4 Evolution of the subcontinental lithospheric mantle). In Section 3.5, we summarize the differences between this experiment
Discussion
The geodynamics of Earth in the Archean is a matter of current debate (O’Neill and Debaille, 2014) and the genesis of the early continental crust remains contentious (Herzberg and Rudnick, 2012, Johnson et al., 2014). Geodynamic modeling of the Archean using higher mantle temperatures and higher internal heat production has shown that subduction was unlikely before the Mesoarchean, which has led to the suggestion that the predominant geodynamic regime prior to the Mesoarchean was stagnant-lid
Conclusions
Results of the 2D petrological–thermomechanical tectono-magmatic numerical experiments developed for conditions appropriate to the hotter early Archean lithosphere demonstrate a variety of tectono-magmatic settings in which felsic melts can be generated from hydrated primitive basaltic crust: (1) delamination and dripping of the lower primitive basaltic crust into the mantle; (2) local thickening of the primitive basaltic crust; and (3) small-scale crustal overturns. Based on the P–T conditions
Acknowledgments
We acknowledge the two journal reviewers, J.F. Moyen and C. Herzberg, for detailed comments that helped us improve the paper during revision. This work was supported by Austrian Science Fund (FWF), Lise Meitner Program, project number M 1559-N29 (E.S.) and by an ETH-grant ETH-37_11-2 (T.G.). Simulations were performed on the ETH-Zurich Brutus cluster.
References (129)
- et al.
Geochemistry and origin of the Early Precambrian crust of north-eastern Minnesota
Geochim. Cosmochim. Acta
(1975) A catalytic delamination-driven model for coupled genesis of Archaean crust and sub-continental lithospheric mantle
Geochim. Cosmochim. Acta
(2006)- et al.
The hunting of the snArc
Precambrian Res.
(2013) Heat loss from the Earth: a constraint on Archean tectonics from the relation between geothermal gradients and the rate of plate production
Earth Planet. Sci. Lett.
(1978)The contribution of metamorphic petrology to understanding lithosphere evolution and geodynamics
Geosci. Front.
(2014)Origin and early growth rate of continents
Precambrian Res.
(1986)Rates of magma emplacement and volcanic output
J. Volcanol. Geotherm. Res.
(1984)- et al.
Earth Planet. Sci. Lett.
(2013) - et al.
Burial and exhumation during Archean sagduction in the East Pilbara Granite–Greenstone Terrane
Earth Planet. Sci. Lett.
(2014) - et al.
Rayleigh-Taylor instabilities from hydration and melting propel “cold plumes” at subduction zones
Earth Planet. Sci. Lett.
(2003)