Heterogeneous Hadean crust with ambient mantle affinity recorded in detrital zircons of the Green Sandstone Bed, South Africa

Abstract

The nature of Earth's earliest crust and the processes by which it formed remain major issues in Precambrian geology. Due to the absence of a rock record older than ∼4.02 Ga, the only direct record of the Hadean is from rare detrital zircon and that largely from a single area: the Jack Hills and Mount Narryer region of Western Australia. Here, we report on the geochemistry of Hadean detrital zircons as old as 4.15 Ga from the newly discovered Green Sandstone Bed in the Barberton greenstone belt, South Africa. We demonstrate that the U-Nb-Sc-Yb systematics of the majority of these Hadean zircons show a mantle affinity as seen in zircon from modern plume-type mantle environments and do not resemble zircon from modern continental or oceanic arcs. The zircon trace element compositions furthermore suggest magma compositions ranging from higher temperature, primitive to lower temperature, and more evolved tonalite-trondhjemite-granodiorite (TTG)-like magmas that experienced some reworking of hydrated crust. We propose that the Hadean parental magmas of the Green Sandstone Bed zircons formed from remelting of mafic, mantle-derived crust that experienced some hydrous input during melting but not from the processes seen in modern arc magmatism.

Keywords: Hadean; crustal evolution; early Earth; zircon.

Fig. 1.

Chondrite-normalized REE abundance of Hadean zircons from the GSB. (A) REE patterns of Hadean Group II show depletion in MREE relative to Group I. (B) Average model melt composition derived using Ti-calibrated Kds (17) compared to known Paleoarchean felsic igneous rocks in the Barberton greenstone belt (–76). Group I Hadean zircons show similar model melt compositions to rhyolite of the 3.55 Ga Theespruit and rhyodacite of the 3.25 Ga Bien Venue Formations; Group II shows a similarly low HREE abundance to the 3.45 Ga Hooggenoeg dacite and the 3.23 Ga Kaap Valley Tonalite. LREEs were not evaluated because their low abundances and the potential influence of minute inclusions makes the LREEs inherently unreliable (16). (C) Representative CL images of zircons from Group I and Group II.

Fig. 2.

(A) PCA in zircon trace element ratios. Principal component 1 accounts for 51.1% and principal component 2 for 16.1% of the variability. (B and C) Zircon trace element plots modified after Grimes et al. (16). (B) U/Yb ratio versus Hf for zircons from the GSB. (C) U/Yb versus Nb/Yb plot with inset showing the effect of open-system (Rayleigh) fractionation of select minerals on the displayed system (16). (D) U/Yb versus Nb/Sc plot. Colored fields represent a global compilation of zircons from different Phanerozoic tectono-magmatic settings. The outer contour line is shown at the 95% level, which represents the amount of the distribution inside the contour. Green circles represent Group I zircons; red circles represent Group II.

Fig. 3.

Ti (ppm) and T^xlln box-and-whisker plots (the ends of the boxes are the upper and lower quartiles, vertical lines mark the median, and horizontal lines extend to the highest and lowest observations) of Hadean zircons of the GSB in comparison to those of zircons derived from the Jack Hills (37), TTGs of the Acasta Gneiss Complex (36), TTGs of the Barberton greenstone belt, mafic and felsic rocks (35), and different tectono-magmatic environments in the Phanerozoic (16). All temperatures were calculated with αSiO₂ and αTiO₂ of 1.

Fig. 4.

Geochemical comparison between Jack Hills (3, 37, 38) (gray dots) and GSB Hadean zircons. (A) U/Yb versus Hf, (B) U/Yb versus Gd/Yb, (C) U/Yb versus Nb/Yb, and (D) Th/U versus Ti (ppm). Fields for Phanerozoic zircons are from Grimes et al. (16).

Fig. 5.

Probability density plots of detrital zircons from (A) the base and the top of the GSB in the Kaapvaal Craton and (B) the Maynard Hills greenstone belt (47), and a representative set of detrital zircon dates from the Jack Hills area (77) in the Yilgarn Craton. Prominent age clusters (colored) were calculated using weighted mean ages and overlap within 2 σ-level.