Contemporaneous mobile- and stagnant-lid tectonics on the Hadean Earth

Sample preparation, imaging and rejection criteria

The zircons in this study came from JH samples of Jack Hills metaconglomerate from the Discovery Outcrop on Eranondoo Hill¹, from outcrops up to 900 m to the east, and from one cross-bedded metasandstone (01JH-36). Samples were reduced to sand size by electropulse disaggregation. Zircons were concentrated hydraulically on a gold table and by heavy liquids before separation of the low magnetic susceptibility zircon fraction by Frantz^48,49. Zircons were hand-picked by binocular microscope, cast into 25.4 mm diameter epoxy mounts with analysis standards, and ground or polished at their approximate mid-sections.

The JH zircons of this study are divided into two suites. Sample mounts JH- and W- (01JH-13(a,b), 01JH-36, 01JH-42, 01JH-47, 01JH-54(a,b), 01JH-60(a,b) and W-74-3 and W-74-4) were extensively studied previously^{1,12,26,50,51} and in this study. Samples 01JH-12 and 03JH-141 were collected 1 m from 01JH-13, 01JH-54, W-74-3, and W-74-4. Zircons in mounts ERC-2 through ERC-10 were separated from 01JH-12 in 2022.

Before mounting, the zircons in mounts ERC-2 to ERC-10 were experimentally heated to 1,100 °C at 0.4 GPa of Ar atmosphere for 6 h in a rapid-quench internally heated gas apparatus to fuse melt inclusions for subsequent study. The zircons in sample mounts with no prefix ERC were not experimentally heated. Differences are observed in levels of retained radiation damage seen by Raman (effective dose) and cathodoluminescence, none of which influence the results reported here. No differences are observed between experimentally annealed zircons and not annealed zircons for age, oxygen isotope ratio or trace elements in the selected low-magnetic-susceptibility zircons with >95% concordant U–Pb ages⁵². This is consistent with many studies of chemically abraded zircons that were experimentally heated to 800–1,100 °C or above⁵³ and further attests to the refractory and retentive nature of crystalline zircons.

Procedures for SIMS analysis of age (U–Pb), oxygen isotope ratio (δ¹⁸O and ¹⁶OH/¹⁶O) and trace element concentrations (REEs, Nb, Sc, Al, P, Ca, Ti, Fe, Y, Hf, Ta, Th and U) have been described elsewhere and are summarized below. A new aspect of these data is the analysis of Nb, Sc and Ta in zircon, which requires higher mass resolution than routinely used in studies with forward-geometry SIMS instruments³⁰.

The polished mid-sections of zircons were imaged optically and by SEM (back-scattered electrons, secondary electrons and cathodoluminescence) before and after SIMS analysis. Pre-SIMS imaging allowed selection of analysis spots to avoid obvious alteration and inclusions, and targeting of sub-domains in zoned zircons. Post-SIMS imaging evaluated whether pits were irregular (that is, placed on inclusions, cracks and alteration zones) and in the selected domain^50,54,55. Data from irregular pits were not considered further.

Zircon analyses were further filtered by composition. Acceptance criteria were conservatively set as follows: >95% concordant U–Pb ages; low concentrations of non-formula elements (<30 µg g⁻¹ Al or Ca; <50 µg g⁻¹ Fe); La < 0.25 µg g⁻¹; Pr_N < 10; (Sm/La)_N > 10; Th/U > 0.1; Ce/Ce* > 3; and for δ¹⁸O, ¹⁶O/¹H¹⁶O < 0.0004 (refs. ^{26,50,55,56,57,58,59}). These zircons all passed the LREE-I test⁵⁵, which is less stringent.

Geochronology

All SIMS analyses were made by CAMECA IMS 1280 in the WiscSIMS Lab, University of Wisconsin–Madison. Zircons in mounts ERC-8, ERC-9 and ERC-10 were surveyed for quick ages in automatic mode. Samples were cleaned and carbon-coated to minimize common lead contamination and a Hyperion-II RF-source produced a 5-nA ¹⁶O₂⁻ beam focused to around 15 μm on the sample surface. Simultaneous analysis of ²⁰⁴Pb⁺, ²⁰⁶Pb⁺, ²⁰⁷Pb⁺ and ⁹²Zr₂O⁺ took about 1.3 min per analysis (Supplementary Table 1). These quick model ages are based on ²⁰⁷Pb/²⁰⁶Pb.

Full U–Pb analyses were measured in this study for selected zircons in the ERC-2 to ERC-10 mounts. The data for JH and W mounts are compiled from earlier sources^1,12,26. The ERC mounts were repolished to remove carbon and coated with Au for full U-Pb analysis. Gold coating (rather than C) yields higher and more stable count rates on Pb, but can include Pb from the Au, which must be removed by pre-sputtering before analysis and corrected for common Pb⁶⁰. The primary beam of ¹⁶O₂⁻ (Hyperion-II RF-source) was accelerated at 10 keV for an impact energy of 23 keV and a spot size of about 20 μm. An oxygen leak was used at a chamber pressure of about 1–2 × 10⁻⁵mbar to improve the Pb yield. Each counting cycle proceeded through the masses: ⁹²Zr₂¹⁶O⁺ [C], ⁹²Zr₂¹⁶O⁺ [L1], 200.5 (blank), simultaneous collection of Pb isotopes ²⁰⁴Pb⁺, ²⁰⁶Pb⁺ and ²⁰⁷Pb⁺, simultaneous collection of ²³⁸U⁺ and ²³⁸U¹⁶O⁺ and ²³⁸U¹⁶O₂⁺ at MRP ~ 8,000. A total of six cycles were measured and the last five were integrated for age calculations. For all sessions except January 2024, the measured values of ²⁰⁶Pb/²³⁸U were corrected based on the ²⁰⁶Pb/²³⁸U compared with ²³⁸U¹⁶O₂/²³⁸U trend as measured on reference zircon 91500 (refs. ^61,62), which was also used to determine and correct the instrumental mass fractionation of Pb isotopes. In January 2024, calibration of U–Pb isotope ratios used zircon reference material M127 (ref. ⁶³), which was also used to determine and correct the instrumental mass fractionation of Pb isotopes. Temora-1, Temora-2 (ref. ⁶⁴), 91500 and M127 were run as secondary reference materials in each session. Measured ²⁰⁴Pb was used to correct for common Pb using the two-stage Pb evolution model in ref. ⁶⁰. Ages are based on ²⁰⁷Pb/²⁰⁶Pb in the >95% concordant analyses.

Oxygen isotope ratio and OH/O

Oxygen isotope ratios and ¹⁶O¹H/¹⁶O (OH/O hereafter) were measured in three sessions for the ERC mounts. The data for 01JH and W mounts are compiled from earlier sources^1,26. Analysis conditions were the same as described previously^26,65,66,67. The primary beam of ¹³³Cs⁺ was accelerated at 10 keV and focused to an approximately 10 μm diameter spot (about 1 μm deep) on the gold-coated sample surface with an impact energy of 20 keV. The ions of ¹⁶O⁻, ¹⁶O¹H⁻ and ¹⁸O⁻ were analysed in multi-collector mode using three Faraday cup detectors. Primary beam currents were about 2 nA. Secondary ion yields for ¹⁶O⁻ varied from 1.43 Gcps nA⁻¹ to 1.57 Gcps nA⁻¹ and were nearly constant within each session. Analyses of zircon with ion yields more than 5% different from bracketing zircon standards are rare and were discarded. Four analyses on the UWZ1 zircon standard (δ¹⁸O = 4.96‰ Vienna Standard Mean Ocean Water (VSMOW); ref. ⁶⁸) were made before and after every 10–15 sample analyses. The average precision for spot-to-spot analyses of δ¹⁸O on standards averages 0.15‰, 0.24‰ and 0.18‰ (2 standard deviations), respectively, for the three sessions. The measured ratios of ¹⁸O to ¹⁶O are corrected for instrumental bias as calculated from bracketing UWZ1 zircon analyses and reported in standard δ¹⁸O notation relative to VSMOW. No correction is necessary for matrix effects because of variable [Hf] in these zircons that mostly range from 7,000 μg g⁻¹ to 12,000 μg g⁻¹ (maximum = 15,723, average = 9,508 ± 1,244 μg g⁻¹ for 410 analyses), which is close to UWZ1 (10,187 μg g⁻¹) or KIM5 (8,960 µg g⁻¹), but substantially higher than 91,500 (6,030 µg g⁻¹; ref. ⁶⁸).

High background-corrected values of OH/O can correlate to erroneous values of δ¹⁸O in altered zircon^26,59,69. This correlation is also reported in ref. ⁷⁰, which corrected OH/O with long-term averages in more heavily damaged JH zircons. Background correction in this study is based on the nominally anhydrous standard (UWZ1) analysed approximately every 2 h, and better describes improving vacuum conditions after each sample change. Values of OH/O > 0.0004 were rare in zircons that passed other acceptance tests and were rejected.

Trace elements

Recent advances facilitate routine and accurate trace-element analysis of zircon by SIMS, including new protocols for high mass-resolution analysis that is required for Nb and Sc³⁰; new zircon reference materials; smaller spot sizes (1–10 μm diameter); and tests to identify altered domains.

Trace and minor elements were measured in five sessions using magnet-scanning automatic-analysis mode by CAMECA IMS 1280 at WiscSIMS: Al, P, Ca, Sc, Ti, Fe, Y, Nb, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, Hf, Ta, Th and U. An additional 12 masses were counted for purposes of calibration, to evaluate interferences and to keep the mass spectrometer tuned. A Hyperion-II RF-source provided the primary ¹⁶O⁻ beam that was accelerated at 10 keV with no energy offset for an impact energy of 23 keV and a beam current of about 5.0 nA. Spot sizes are about 15 μm in diameter. Secondary ions were detected with the axial ETP electron multiplier, except major elements (²⁸Si, ⁹⁰Zr), which were counted by the Faraday cup detector.

The mass spectrometer was tuned with an MRP ≈ 12,500, as measured on ⁹³Nb. Quantification of peak shapes and direct measurement of ⁹²ZrH⁺ allow for the estimation of backgrounds for limiting interferences (⁹³Nb⁺ with ⁹²ZrH⁺, and ⁴⁵Sc⁺ with ⁹⁰Zr²⁺; ref. ³⁰). ⁹²ZrH⁺ signals are dependent on several factors, including chamber pressure and the hydration state of the target zircon domain. For analyses in this study, ⁹²ZrH⁺ contributions to ⁹³Nb⁺ are a maximum of 0.17 µg g⁻¹, and average 0.0079 µg g⁻¹; relative signal levels average <0.2% of the peak value, with a maximum of 2%. ⁹⁰Zr²⁺ is not measured directly, as Zr²⁺ is expected to be relatively consistent from analysis to analysis. ⁹⁰Zr²⁺ backgrounds on ⁴⁵Sc⁺ are calculated to be to be a maximum of <0.009 µg g⁻¹, with an average 0.007 µg g⁻¹, and have relative contributions averaging 0.04%, with a maximum of 1.1%.

At higher MRP, accurate mass calibration is especially sensitive to small changes in magnetic field and is aided by maintaining precise timings of magnet scans, and monitoring or adjustment of the mass calibration. In this study, tuning was maintained for several days with a new monitoring and correction routine that tracks and makes regular ppm-level adjustments to the magnetic field after every analysis. This routine monitors the position of five major-element peaks (²⁸Si⁺, ⁹⁰Zr⁺, ⁹⁰Zr¹⁶O₂⁺, ⁹⁰Zr₂⁺, and ⁹⁰Zr₂¹⁶O₃⁺) with sensitive measurements on the peak and its flank and adjusts the magnet calibration to keep both reference and interpolated masses centered³⁰.

Each analysis included a 30-s pre-sputter, centring of the secondary beam, and five cycles of counting from low to high mass. Count rates change more rapidly early in each analysis. Thus, the last four cycles are normalized to ²⁸Si (on a cycle-by-cycle basis) and averaged for use in calculating concentrations.

Calculation of trace element concentrations is based on the conversion of normalized count rates between zircon and the glass standard NIST610 that was analysed regularly throughout each session. For most elements, this calculation incorporates a correction factor based on the comparison of trace element concentrations measured in zircon reference materials. Five reference materials were analysed: NIST 610 glass⁷¹, and four zircons, 91500 (refs. ^62,72), MAD-559 (ref. ⁷²), M127 and GZ7 (ref. ⁷³) (Supplementary Table 2). Oxygen isotope zircon standards (UWZ1 and KIM5) were also analysed to test instrument conditions, but these zircons are not sufficiently homogeneous in trace elements to be used for calibration. We note that ref. ¹⁴ calibrated SIMS data with a combination of zircon reference materials CZ3 and MAD-1. We use MAD559, which was intercalibrated with these materials in ref. ⁷². In this study, we used the original discriminants¹⁴. This is consistent with published studies for BGSB¹³ and other zircons^13,41,45. A previous study⁴⁰ suggests revision of ref. ¹⁴ Sc/Yb and U/Nb discriminants. These differences are small and within analytical uncertainty.

Each analysis in our EarthChem dataset has calculated trace element concentrations in µg g⁻¹, and for the lanthanides, the chondrite normalized values are also shown⁷⁴. The measured compositions of U and Th are used in this study in agreement with the data in ref. ¹⁴ and published BGSB data¹³. If compositions of U were corrected for radioactive decay since the time of crystallization of Hadean zircon, they would be nearly twice as high, which would move U/Nb ratios away from the OI field if discriminant boundaries are not similarly adjusted, and making more JH zircons appear to plot in the field of continental crust.

Age distribution in sample 01JH-12

The percentage of Hadean zircons separated from different samples of JH metasediment in different studies varies greatly. It is important to recognize that no suite of detrital zircons is necessarily representative of their source rocks. Sorting occurs in nature because of the differences in erosion and possible selective destruction of metamict or otherwise damaged grains. Sorting also occurs in the lab because of heavy mineral concentration, magnetic sorting, hand picking, imaging and selection of analysis spots. To obtain the distribution of ages for selected zircons from epoxy mounts in this study, all zircons in 3 of the 19 mounts examined (ERC-8, ERC-9 and ERC-10; from sample 01JH-12 at the Discovery Outcrop) were dated by quick analyses (Extended Data Fig. 1a, n = 2,843). As a test of accuracy for the quick ages, a subset of these zircons, mostly Hadean, were selected for the full U–Pb analysis. Of the 125 zircons that are over 95% concordant, the U–Pb ages are on average only 4 Myr different from quick ages, including some grains with large age differences that may be zoned (Extended Data Fig. 1b).

As in other studies of JH zircons^1,2,75, most of the zircons from 01JH-12 are Archean. The dominant population has an average age of 3.4 Gyr (Extended Data Fig. 1a), similar to gneisses surrounding the JH. Curiously, there are few zircons dated to between 3.7 Ga, the age of the oldest known rocks in Australia, and 3.9 Ga. The Hadean zircons represent a different subset with a peak at 4 Ga extending to 4.4 Ga for concordant U–Pb ages. Most of the Hadean ages were verified by a full U–Pb analysis. Of the 2,843 quick ages in mounts ERC-8, ERC-9 and ERC-10, there are 300 older than 3.9 Gyr and 218 older than 4.0 Gyr (10.6% and 7.8%, respectively). Moreover, a small number of zircons with quick ages <4.0 Gyr are shown by full U–Pb analyses to be older than 4.0 Gyr, indicating that the actual percentage of Hadean zircons in the sample is 8.8%.

The reported percentages of Hadean (>4.0 Ga) ages in other studies of the JH zircons range from about 3 to 12% among concordant zircons^{1,2,12,25,26,27,46,75,76}. Even at 3%, this is a notably higher percentage than at any other known locality^77,78. Ages younger than 3.0 Gyr, the putative age of sedimentation, have been reported from JH metasediments and could result from contamination in the field⁷⁹ or otherwise unrecognized differences in lithology²⁵. These young ages were not encountered in this study (with one exception) and are ignored in other databases.

Geodynamic modelling

Previous studies^37,44 applied the mantle convection code StagYY⁸⁰ to model the Hadean and Eoarchean Earth in a two-dimensional spherical annulus geometry⁸¹ (Extended Data Fig. 2). These models incorporate pressure- and temperature-dependent water solubility maps for different mantle minerals⁸², composite rheology (diffusion creep and dislocation creep proxy), the effect of water on the density of mantle minerals and the frictional strength of the oceanic lithosphere of early Earth to estimate the production of tonalite, trondhjemite and granodiorite magmas and the evolution of trace elements including U and Nb.