How continental keels and cratons may have formed

There is Byzantine ring to the word craton: hardly surprising as it stems from the Greek kratos meaning ‘might’ or ‘strength’. Yes, the ancient cores of the continents were well named, for they are mighty. Some continents, such as Africa, have several of them: probably relics of very ancient supercontinents that have split and spread again and again. Cratons overlie what are almost literally the ‘keels’ of continents. Unlike other mantle lithosphere beneath continental crust (150 km on average) cratonic lithosphere extends down to 350 km and is rigid. Upper mantle rocks at that depth elsewhere are mechanically weaker and constitute the asthenosphere. Geologists only have evidence from the near-surface on which to base ideas of how cratons formed. Their exposed rocks are always Precambrian in age, from 1.5 to 3.5 billion years old, though in some cases they are covered by a thin veneer of later sedimentary rocks that show little sign of deformation. No cratons formed after the Palaeoproterozoic and they are the main repositories of Archaean rock. Their crust is thicker than elsewhere and dominated at the surface by crystalline rocks of roughly granitic composition. Cratons have the lowest amount of heat flowing out from the Earth’s interior; i.e. heat produced by the decay of long-lived radioactive isotopes of uranium, thorium and potassium. This relative coolness provides an explanation for the rigidity of cratons relative to younger continental lithosphere. Because granitic rocks are well-endowed with heat-producing isotopes, the implication of low heat flow is that the deeper parts of the crust are strongly depleted in them. As a result the deep mantle in cratonic keels is at higher pressure and lower temperature than elsewhere beneath the continental surface. Ideal conditions for the formation of diamonds in mantle rock, so that cratonic keels are their main source – they get to the surface in magma pipes when small amounts of partial melting take place in the lithospheric mantle.

The low heat flow through cratons beckons the idea that the heat-producing elements U, Th and K were at some stage driven from depth. An attractive hypothesis is that they were carried in low-density granitic magmas formed by partial melting of mantle lithosphere during the Precambrian that rose to form continental crust. Yet there is an abundance of younger granite plutons that are associated with thinner continental lithosphere. This seeming paradox suggests different kinds of magmagenesis and tectonics during the early Precambrian. Russian and Australian geoscientists have proposed an ingenious explanation (Perchuk, A.L. et al. 2020. Building cratonic keels in Precambrian plate tectonics. Nature, v. 586, p. 395-401; DOI: 10.1038/s41586-020-2806-7). The key to their hypothesis lies in the 2-layered nature of mantle keels beneath cratons, as revealed by seismic studies. Modelling of the data suggests that the layering resulted from different degrees of partial melting in the upper mantle during Precambrian subduction.

Development of a cratonic keel from melt-depleted lithospheric mantle during early Precambrian subduction. Mantle temperature is 250°C higher than it is today. The oceanic lithosphere being subducted in (a) has become a series of stagnant slabs in (b) (credit: Perchuk et al.; Fig. 2)

Perchuk et al. suggest that high degrees of partial melting of mantle associated with subduction zones produced the bulk of magma that formed the Archaean and Palaeoproterozoic crust. This helps explain large differences between the bulk compositions of ancient and more recent continental crust, which involves less melting. The residue left by high degrees of melting of mantle rock in the early Precambrian would have had a lower density than the rest of the mantle. While older oceanic crust at ancient subduction zones would be transformed to a state denser than the mantle as a whole and thus able to sink, this depleted lithospheric mantle would not. In its hot ductile state following partial melting, this mantle would be ‘peeled’ from the associated oceanic crust to be emplaced below. The figure shows one of several outcomes of a complex magmatic-thermomechanical model ‘driven’ by assumed Archaean conditions in the upper mantle and lithosphere An excellent summary of modern ideas on the start of plate tectonics and evolution of the continents is given by:Hawkesworth, C.J., Cawood, P.A. & Dhuime, B. 2020. The evolution of the continental crust and the onset of plate tectonics. In Topic: The early Earth crust and its formation, Frontiers in Earth Sciences; DOI: 10.3389/feart.2020.00326

An Early Archaean Waterworld

In Earth-logs you may have come across the uses of oxygen isotopes, mainly in connection with their variations in the fossils of marine organisms and in ice cores. The relative proportion of the ‘heavy’ 18O isotope to the ‘light’ 16O, expressed by δ18O, is a measure of the degree of fractionation between these isotopes under different temperature conditions when water evaporates. What happens is that H216O, in which the lighter isotope is bound up, slightly more easily evaporates thus enriching the remaining liquid water in H218O. As a result the greater the temperature of surface water and the more of evaporates, the higher is its δ18O value. Shells that benthonic (surface-dwelling) organism secrete are made mainly of the mineral calcite (CaCO3). Their formation involves extracting dissolved calcium ions and CO2 plus an extra oxygen from the water itself, as calcite’s formula suggests. So plankton shells fossilised  in ocean-floor sediments carry the δ18O and thus a temperature signal of surface water at the place and time in which they lived. Yet this signal is contaminated with another signal: that of the amount of water evaporated from the ocean surface (with lowered  δ18O) that has ended up falling as snow and then becoming trapped in continental ice sheets. The two can be separated using the δ18O found in shells of bottom-dwelling (benthonic) organisms, because deep ocean water maintains a similar low temperature at all time (about 2°C). Benthonic δ18O is the main guide to the changing volume of continental ice throughout the last 30 million year or so. This ingenious approach, developed about 50 years ago, has become the key to understanding past climate changes as reflected in records of ice volume and ocean surface temperature. Yet these two factors are not the only ones at work on marine oxygen isotopes.

Artistic impression of the Early Archaean Earth dominated by oceans (Credit:

When rainwater flows across the land, clays in the soil formed by weathering of crystalline rocks preferentially extract 18O and thus leave their own δ18O mark in ocean water. This has little, if any, effect on the use of δ18O to track past climate change, simply because the extent of the continents hasn’t changed much over the last 2 billion years or so. Likewise, the geological record over that period clearly indicates that rain, wet soil and water flowing across the land have all continued somewhere or other, irrespective of climate. However, one of the thorny issues in Earth science concerns changes of the area of continents in the very long term. They are suspected but difficult to tie down. Benjamin Johnson of the University of Colorado and Boswell Wing of Iowa State University, USA, have closely examined oxygen isotopes in 3.24 billion-year old rocks from a relic of Palaeoarchaean ocean crust from the Pilbara district of Western Australia that shows pervasive evidence of alteration by hot circulating ocean water (Johnson, B.W. & Wing, B.A. 2020. Limited Archaean continental emergence reflected in an early Archaean 18O-enriched ocean. Nature Geoscience, v. 13, p. 243-248; DOI: 10.1038/s41561-020-0538-9). Interestingly, apart from the composition of the lavas, the altered rocks look just the same as much more recent examples of such ophiolites.

The study used many samples taken from the base to the top of the ophiolite along some 20 traverses across its outcrop. Overall the isotopic analyses suggested that the circulating water responsible for the hydrothermal alteration 3.2 Ga ago was much more enriched in 18O than is modern ocean water. The authors’ favoured explanation is that much less continental crust was exposed above sea level during the Palaeoarchaean Era than in later times and so far less clay was around on land. That does not necessarily imply that less continental crust existed at that time compared with the Archaean during the following 700 Ma , merely that the continental ‘freeboard’ was so low that only a few islands emerged above the waves. By the end of the Archaean 2.5 Ga ago the authors estimate that oceanic δ18O had decreased to approximately modern levels. This they attribute to a steady increase in weathering of the emerging continental landmasses and the extraction of 18O into new, clay-rich soils as the continents emerged above sea level. How this scenario of a ‘drowned’ world developed is not discussed. One possibility is that the average depth of the oceans then was considerably less than it was in later times: i.e. sea level stood higher because the volume available to contain ocean water was less. One possible explanation for that and the subsequent change in oxygen isotopes might be a transition during the later Archaean Eon into modern-style plate tectonics. The resulting steep subduction forms deep trench systems able to ‘hold’ more water. Prior to that faster production of oceanic crust resulted in what are now the ocean abyssal plains being buoyed up by warmer young crust that extended beneath them. Today they average around 4000 m deep, thanks to the increased density of cooled crust, and account for a large proportion of the volume of modern ocean basins.

The oldest impact structure

Ilulisat Isfjord
Ilulisat Grenland (credit: Wikipedia)

Various lines of evidence, such as sedimentary deposits of glass spherules and shocked minerals or signs of unusual isotopic chemistry (see Ejecta from the Sudbury impact and Evidence builds for major impacts in Early Archaean in EPN April 2005 and August 2002) point to the predicted intensity of meteorite or comet bombardment of the early Earth, and evidence is growing for some events that had global effects. Yet no actual impact sites from the Archaean Eon have been found, until recently. That is not entirely unexpected because erosion during the last few billion years will have removed all trace of the characteristic surface craters. But perhaps there is cryptic evidence in Archaean terrains for the deeper influence of impacts: after all, the depth of penetration of large meteoritic ‘missiles’ would have been of a similar order to their diameter where shock structures in minerals would slowly anneal and impact-generated melts would crystallise slowly enough to masquerade as plutonic igneous rocks.

Close to the Arctic Circle in SW Greenland Archaean gneisses are associated with a roughly 200 km wide geomagnetic anomaly and regionally curvilinear features that suggest a series of concentric closed structures over a 100 km diameter area (Garde, A.A. et al. 2012. Searching for giant, ancient impact structures on Earth: The Mesoarchaean Maniitsoq structure, West Greenland. Earth and Planetary Science Letters, v.  337, p. 197-210). Adam Garde and colleagues from the Greenland Geological Survey, Cardiff University UK and Lund University Sweden focused on the central part of these anomalies where gneisses are extensively brecciated with signs of annealed shock-induced lamellae in quartz, feldspar melting and fluidization of highly comminuted mylonites. They ascribe this assemblage of features on a variety of scales to the effects of a major meteorite impact on 25 km deep continental crust. The metamorphic complex contains the famous Amitsoq Gneisses that once had the status of the world’s oldest rocks at around 3.8 Ga, but is dominated by migmatites formed around 3.1 Ga that are akin to the Nuuk Gneisses from further south.

The possible signs of a deeply penetrating impact are cut through by small ultramafic intrusions, zircons from which yield 207Pb/206Pb ages between 3.01 and 2.98 Ma, confirming the structures’ Mesoarchaean age. An interesting and unanswered question concerns the origin of these magmas together with marginally younger, voluminous granites. Were the ultramafic magmas generated by high degrees of partial melting of mantle as a result of the immense energy of impact?  Having temperatures well above those of basaltic melts, could the ultramafic intrusions in turn have induced crustal melting within the depths of a large impact basin?