The nature of the tectonic processes that shaped the early Earth remain unresolved, hampered not only by the sparse Early Archaean crustal rock record, but also by the dearth of tangible mantle samples (e.g., xenoliths and diamonds) older than 3 Ga. Investigating the Archaean mantle provides a complementary foil to the knowledge gleaned from the early Earth crust, and can be used to trace the onset of crustal recycling, but also to evaluate the secular evolution of Earth’s mantle regarding its temperature and composition including redox state.

We have conducted an in-situ carbon and nitrogen isotope study of “confirmed” Archaean diamonds from the 3.0 – 2.8 Ga Witwatersrand Supergroup of the Kaapvaal craton in South Africa [1]. While the absolute formation age of the placer diamonds is unknown, nitrogen aggregation suggests diamond residence within the upper mantle for 10 - 400 Myr. Coupled with the depositional age of the Archaean basin, the Witwatersrand diamonds may have formed in the mantle as early as 3.5 Ga, before their transport via kimberlite-like magmatism to Earth’s surface during formation of the Kaapvaal craton.

The d¹⁵N values of 0.5 to +2.7 ‰ determined for the Witwatersrand diamonds are higher than both the ancient and modern mantle (-5 ‰), and overlap with positive d¹⁵N values shown by >3 Ga old Kaapvaal sedimentary rocks. The diamond carbon isotope ratios (d¹³C of -5.7 to -3 ‰) are mantle-like, but increases in d¹³C values from core to rim suggest that the Witwatersrand diamonds formed from relatively oxidised fluids containing CO₂ rather than CH₄. It follows that oxidised CHO-fluids containing recycled crustal nitrogen were present in the upper mantle possibly prior to 3.5 Ga. This observation suggests operation of subduction-style tectonics during the inception of craton formation in the Eo- to Palaeoarchaean. It also implies that the Early Archaean upper mantle was not more reducing than at the present, in alignment with new evidence for an oxidised CO₂-rich early Earth atmosphere created by mantle outgassing.

[1] Smart KA, Tappe S, Stern RA, Webb SJ and Ashwal LD. 2016. Early Archaean tectonics and mantle redox recorded in Witwatersrand diamonds. Nature Geoscience, v. 9, p. 255–259.

The oldest volcanic rocks exposed on Malekula Island, now belonging to the New Hebrides Island Arc, formed in the Upper Eocene to Middle Miocene Vitiaz Island Arc, Southwest Pacific. They are thought to have formed contemporaneously with Fiji and the Izu-Bonin-Mariana (IBM) arc during westward subduction initiation of the Pacific beneath the Indo-Australian Plate [e.g., 1]. To test this hypothesis with regard to the mantle source compositions and contributions from the subducting slab, we provide major- and trace element data combined with Hf, Nd, and Pb isotopes for twenty-seven volcanic rocks of Malekula Island. Our results show that Malekula lavas display similar magma types, i.e., boninite-series rocks, island arc basalts, and MORB-type tholeiites, to the earliest volcanic rocks of Fiji and the IBM arc resembling the sequential stratigraphy of the IBM system [2], rather than the interlayered stratigraphy of early arc rocks on Fiji [3].

Moreover, Malekula lavas display a change in Hf-Nd isotope composition from isotopically ‘Indian’, similar to the IBM arc [4, 5], to mainly ‘Pacific’, like on Fiji [3]. We interpret this progressive change in mantle source composition to reflect the propagation of ‘Pacific’ South Fiji Basin spreading into the Vitiaz Arc. Hence, the Malekula lava succession provides a link between subduction initiation in the Northwest and Southwest Pacific.

[1] Hall (2002) J Asian Earth Sci 20. [2] Ishizuka et al. (2011) EPSL 306. [3] Todd et al. (2012) EPSL 335-336. [4] Reagan et al. (2010) G³11(3). [5] Li et al. (2019) EPSL 518.

Phase changes in the mantle have long been known to play a major role for convection in a one-component mantle. When considering cases with depleted ambient upper mantle and upwelling mantle either chemically or mechanically enriched with basaltic crust, very complex density-difference histories are possible for a wide range of realistic temperature-composition scenarios. We explore the ascent of enriched mantle plumes in ambient mantle using combined thermodynamic and themomechanical modelling. Plumes are unlikely to feel a blocking effect from the negative Clapeyron slope of the 660 phase transition due to excessive buoyancy in the uppermost lower mantle. Hot plumes cross the phases transition at temperatures above the negative slope segment and are even promoted. Instead, they may stall and spread in the upper mantle transition zone for significant periods of time, as this depth is characterized by negative thermal expansion for mantle compositions at elevated temperatures. With time, both, the cooling plume and the heating ambient mantle experience density reduction and secondary plumes can spawn from that domain. These secondary plumes may show large lateral offsets from the deep plume stem and show complex and divers geochemical signatures.

We present phase diagrams of variously enriched and depleted mantle rocks down to 800 kilometers depth and explore density as the parameter governing convection and compositional stratification. Some results are surprising and not all are included in present concepts and models:

(1) Primitive and enriched mantle compositions are buoyant in the uppermost lower mantle compared to depleted mantle, especially, when they are warmer, but also at identical temperatures. Hence, if the upper mantle is depleted compared to the lower, a petrological lower-upper-mantle boundary (LUMB) can be expected several tens of kilometers below the seismic one.

(2) Depleted compositions show the slope-break of the 660 phase transitions at higher temperatures. Hence, the uppermost lower mantle would be an excellent trap for very hot depleted mantle, which could be relevant for komatiite generation.

(3) Primitive and enriched compositions experience negative thermal expansion at high temperatures in the upper MTZ, i.e. they display a density minimum at slightly elevated temperatures. The dynamic consequences for plume rise are enormous and explored in a complementary contribution (Vesterholt and Nagel).

The key phase for effects above is garnet, which (1) is stable in the uppermost lower mantle, (2) relatively dens in the upper, but buoyant in the lower mantle, and (3) may become more abundant with temperature. Depending on bulk rock composition, garnet is stable in the uppermost 70-150 kilometers of the lower mantle causing a reversal of the expected density-order in that depth interval. Our present work includes studying seismic footprints of stratification scenarios.