Introduction

The Earth's earliest history is poorly known because of the lack of rocks from its first 500 m.y. (i.e., >4050 Ma). The mafic rocks from the Nuvvuagittuq greenstone belt of the Superior Craton, Canada, produced a short-decay Nd isotope isochron age of ~4300 Ma1. However, the 146Sm-142Nd values might represent early mantle inheritance and the rock age may be as young as ~3800 Ma2. Zircon (ZrSiO4), due to its robust chemical and physical features, may be the only common accessory mineral in continental rocks that could survive the extreme conditions of Earth's early evolution. Rare Hadean-Early Archean zircons are important for understanding the formation of Earth's protocrust and ocean3,4, the characteristics of primordial mantle5 and continental weathering6. Although still debated7,8,9, a number of isotopic observations have suggested that the Hadean Earth may have had a relatively stable, basaltic crustal lid, without plate recycling analogous to modern-style plate tectonics5,10,11,12,13.

It should be noted that our knowledge of early earth crustal evolution has mostly been based on studies of zircons found in Archean metasedimentary rocks in Western Australia (dominantly the Jack Hills)3,4,14,15,16,17. Is this an accident of preservation or was this area unique in the Hadean eon? We are aware of only four > 4050 Ma zircons analyzed by SIMS (secondary ion mass spectrometer) that have been reported as detritus and xenocrysts outside Western Australia: one in the Itsaq Gneiss complex of West Greenland (4079 ± 18 Ma)18, one in the Acasta Gneiss of Northwest Territories, Canada (4189 ± 46 Ma)19 and two in China (one in western Tibet, 4103 ± 4 Ma; and one in the North Qinling Belt, 4080 ± 9 Ma)20,21. There is only one in situ ion microprobe analysis for each of these four zircons and, although they show concordant U-Th-Pb isotopic ages (concordance > 90%), these ages have not been reproduced. These rare survivors from outside Western Australia need careful and comprehensive investigations by a wide range of methods to establish their origins and to help understand the diversity of the earliest crustal evolution of the Earth. In this work, we report SIMS analyses of two > 4050 Ma zircon xenocrysts from one quartzite sample in the Cathaysia Block of southern China. Their oxygen isotope ratios and trace elements (especially Ti concentrations) provide new evidence for the diversity of earliest continental crust.

Geological setting

Southern China (i.e., the South China Block) is composed of the Yangtze Block to the northwest and the Cathaysia Block to the southeast (Fig. 1a). The Cathaysia Block is dominated by Early Paleozoic metamorphic rocks and Mesozoic granitoids and volcanic-sedimentary rocks, with minor Precambrian rocks (including Neoproterozoic 970–750 Myr volcanic-sedimentary rocks and rare Paleoproterozoic gneissic metamorphic rocks). Zircon crystals of this study were separated from a quartzite within the Longquan Group paragneiss that was originally deposited after 700 Ma and metamorphosed at ~450 Ma at the Longquan area of southwestern Zhejiang Province in the Cathaysia Block (Fig. 1b, Supplementary Figs. DR1–2).

Figure 1
figure 1

Geological sketch map showing the sampling location of the two ancient zircon grains.

(a) Southern China composed of the Yangtze and Cathaysia blocks and the Jiangnan orogen between them; (b) Longquan area, southwestern Zhejiang Province. This map image was made by the co-authors Jiang Yang and Xing Guangfu according to their geological survey work, using the software Mapgis and CorelDraw.

Full size image

Results

Two zircon grains (#8 and #123) with 207Pb/206Pb ages older than 4050 Ma were found among 235 U-Pb analyses on 214 zircons from the quartzite sample. Grain #8 shows complicated zoning in CL with an oscillatory-zoned magmatic core surrounded by an intermediate CL-intensity unzoned metamorphic mantle (Th/U ≤ 0.05) and two main outermost unzoned overgrowths (one dark and one bright in CL; Fig. 2a; Supplementary Fig. DR3). In addition, there is a bright unzoned domain in the lower right of the core, showing crosscutting relations with other domains of the core in CL (Fig. 2a; Fig. DR3), possibly representing an altered domain. This is supported by the low Th/U (0.06) and high δ18O (see below). In contrast, grain #123 shows a simple texture in CL, with dark and bright oscillatory zones surrounding an unzoned (possibly radiation damaged) core (Fig. 2b; Fig. DR3).

Figure 2
figure 2

U-Pb Concordia plots for the two ancient zircons.

The CL image (before the second session of U-Pb dating) and age histogram for each grain is also indicated for each grain as insets. The gray filled columns in the lower right insets represent the analyses with concordance > 88%, while the open ones are <88% concordance analyses. Error ellipses are shown in 1σ.

Full size image

The results of the two dating sessions are consistent, suggesting the core (pit #s U8.1, U8.8, U8.10) and metamorphic mantle (#s U8.3, U8.7) of grain #8 are nearly concordant (≥95% concordance except #U8.10; Table 1) and formed in the ranges 4150–4100 Ma and 4070–4060 Ma respectively (Fig. 2a). The altered domain of grain #8 core gives a 207Pb/206Pb age of 4121 ± 7 Ma (1σ; 98% concordance), indistinguishable from other core analyses. In addition, the two outermost pits in the overgrowths of this grain were dated at 3843 ± 7 Ma (1σ) and 3679 ± 6 Ma (1σ) respectively (#U8.6 and #U8.9 in Fig. 2a; Fig. DR3m). For grain #123, nine analyses (5 of 9 with concordance > 89%) from the oscillatory region show 207Pb/206Pb ages in the range 4130–4030 Ma (Table 1; Fig. 2b) and the unzoned core gives a younger 207Pb/206Pb age of 3956 ± 8 Ma (1σ) (82% concordance; #U123.7; Table 1; Fig. 2b; Fig. DR3n).

Table 1 SHRIMP U-Th-Pb isotopic data for two old zircon grains of southern China
Full size table

Four oxygen isotope analyses in the concordant domain of the core of grain #8 give identical results (Fig. DR3g; Supplementary Table S1), yielding an average δ18O of 7.2 ± 0.2‰ (n = 4, 2SD). The metamorphic mantle of grain #8 yielded an average δ18O of 8.0 ± 0.7‰ (n = 4, 2SD), which is similar to the altered domain in the core (δ18O = 7.9 ± 0.2‰, #O8.3; Table S1). The ca. 3840 and 3780 Ma outermost overgrowths yield similar δ18O values (7.6‰ vs. 7.4‰; Table S1; #O8.9, #O8.11 in Fig. DR3g). In contrast, grain #123 shows consistent and mantle-like oxygen isotope ratios in its magmatic zones (δ18O = 5.9 ± 0.2‰, 2SD, n = 6) and the unzoned damaged core gives similar δ18O of 5.6 ± 0.2‰ (#O123.1; Table S1; Fig. DR3h).

The magmatic areas (including the core of #8 and zoned domains of #123) of both grains show a positive slope for chondrite-normalized rare earth element (REE) profiles with significant positive Ce anomalies and negative Eu anomalies (Supplementary Table S2; Fig. 3). They are geochemically similar to pristine continental zircons on Earth in their high (Sm/La)N and U/Yb ratios (Fig. 4). Two analyses from the core of grain #8 showed the highest [Ti] (Ti concentration) (51.5 and 54.0 ppm). The zoned domains of #123 gave relatively low [Ti] (19.8 ppm for dark zonation and 14.7 ppm for light zonation in CL; Table S2). The altered domain and the metamorphic mantle/rim of grain #8 show similar REE patterns, with relatively low REE abundances and weakly negative Eu anomalies (Fig. 3). They also show similar [Ti] (15.3 ppm versus 23.0–18.8 ppm; Table S2).

Figure 3
figure 3

Rare earth element distribution patterns for the two zircons.

Chondrite values are from McDonough and Sun55.

Full size image
Figure 4
figure 4

(Sm/La)N versus La (a) and U/Yb versus Y (b) plots for the analyses from the two old zircons of southern China.

The areas for continental zircons, modern ocean crust zircons, continental granitoid zircons, kimberlite zircons, porous zircons, Jack Hills zircons and hydrothermal zircons are recited from Bouvier et al.52.

Full size image

Discussion

All of the analyses (n = 18) of the two U-Pb sessions show 207Pb/206Pb ages older than 3950 Ma, with the exception of the above-mentioned two outermost overgrowths of grain #8 (#U8.6 and #U8.9 in Fig. 2a). Twelve of the eighteen > 3950 Ma analyses show concordance better than 89% (Table 1). Thus the 207Pb/206Pb ages are not significantly affected by Pb mobilization22 that might lead to anomalously lower or elevated 207Pb/206Pb ages in some parts of a zircon at the expense of other domains. Four analyses (#s U8.1, U8.4, U8.6 and 8.7) show reverse discordance (Fig. 2a). Most of these analyses are from or overlap the low-Th/U metamorphic mantle of grain #8. The reverse discordance may have resulted from localized Pb mobility within partly radiation damaged zircons. The consistency of 207Pb/206Pb ages and good concordance of the analyses suggest that 207Pb/206Pb ages with uncertainties lower than 15 m.y. (1σ) give reliable ages that are the oldest known for zircons from southern China.

High-grade regional metamorphism likely accompanied the formation of the earliest granitic rocks on Earth due to heating from the underplating mantle magmas and high geothermal gradient at that time, although radiogenic heat production is also suggested as a mechanism for crustal melting11. However, such metamorphic rocks have not been identified and may not have survived. Metamorphic zircon can potentially provide a record of the early Earth metamorphism and help understand the crustal evolution of Earth.

Metamorphic zircons can commonly be distinguished from igneous zircons by their low Th/U ratios (generally <0.0723), the absence of oscillatory zonation in CL24 and weak to absent negative Eu anomalies23. The magmatic core of grain #8 is surrounded by a mantle with intermediate CL-intensity that is unzoned (Fig. DR3). Four analyses (trace elements #TE8.3, #TE8.4 and #U8.3, #U8.7) in the mantle all gave Th/U ratios ≤ 0.05 (Table 1 and Table S2), suggesting that the unzoned layer (i.e. the mantle of grain #8) is metamorphic in origin. Hadean to Early Archean analyses with low Th/U (<0.07) ratios have been sporadically reported in detrital zircons from Jack Hills. Cavosie et al.25 reported one analysis (207Pb/206Pb age of 4062 ± 10 Ma, #22-2 of sample 01JH65) with Th/U ratio of 0.04. Unfortunately, this grain was removed during later polishing. Similarly, Harrison et al.9 showed one low Th/U (0.05) analysis (207Pb/206Pb age of 4046 ± 12 Ma, #RSES43-5.7) without detailed oxygen isotopes and CL images. The 4070–4060 Ma metamorphic mantle surrounding the magmatic core of grain #8 shows low Th/U ratio (average 0.04; #U8.7, #TE8.3, #TE8.4; Table 1 and Table S2). Four oxygen isotope analyses (#s O8.1, O8.5, O8.6 and O8.8) in this metamorphic zone gave δ18O ranging from 7.6‰ to 8.3‰ (Table S1), with a mean value of 8.0 ± 0.7‰, which is a little higher than that (7.2 ± 0.2‰) of the magmatic core. The slightly elevated oxygen isotope ratios suggest the existence of 18O-enriched fluids that sourced from low-temperature supracrustal processes. It possibly represents the oldest known metamorphic event on Earth documented by zircon accompanied by reliable CL images, oxygen isotope and trace element analyses.

The titanium-in-zircon thermometer26,27 (also see the equation (1) in the supplementary files) has been widely applied to magmatic zircons. Although the meaning of calculated temperatures and the mechanism of Ti in zircon are still unclear28,29,30,31,32, comparison based on same assumptions can be an effective way to evaluate early crustal evolution33. Most of the Jack Hills analyses (free of cracks) contain less than 20 ppm titanium based on published data (Fig. 5), yielding an average calculated temperature of 697 ± 47°C (1SD, n = 50; Table S3) assuming unit activities of TiO2 and SiO2 and applying no pressure correction (same assumption below). The mean value is consistent with the result (696 ± 33°C) by Watson and Harrison26 and their updated value (682 ± 26°C)4. A few Jack Hills zircons also show temperatures higher than 800°C due to their high [Ti]34, but no detailed CL and/or BSE images were provided along with [Ti] to help evaluate whether the [Ti] analyses had been affected by the existences of micro- cracks and/or inclusions8,34,35,36.

Figure 5
figure 5

Oxygen fugacity (fO2) versus Ti concentration and calculated temperature by the titanium in zircon thermometer26,40.

This plot is made based on a compilation (Table S5) of seventy-three >3800 Ma zircons. The temperatures were calculated assuming unit activities of TiO2 and SiO2 and no pressure correction. Reaction curves of fO2 buffer assemblages (HM, hematite–magnetite; NNO, Ni–NiO; FMQ, quartz–fayalite–magnetite) and the boundary between SO42− and S2− predominance in silicate magmas (dashed line) are from Chou42 and Wallace and Carmichael56 respectively. The black dashed curve represents the oxygen fugacity shown by the two average analyses from the magmatic zonation of the two grains.

Full size image

The calculated low temperature for Jack Hills zircons was suggested to correspond to the temperature of wet minimum melting in present-day crust26,34. However, two analyses (#TE8.1 and #TE8.5) of the magmatic core of grain #8 give consistent [Ti], with a mean value of 53 ± 3.4 ppm, corresponding to a temperature of 910°C by titanium-in-zircon thermometer. This is the highest reported value for a Hadean zircon from Earth (Fig. 5) with detailed CL and oxygen isotopes and it may also represent one of the highest values of [Ti] in terrestrial zircons from any time period29,32,35,37. Interestingly, the metamorphic mantle of grain #8 and the analyses in grain #123 show uncorrected Ti-in-zircon temperatures (776–820°C; Table S2) higher than many of the published Hadean zircons from Jack Hills. Further imaging has shown that no tiny cracks occur in the analyzing pits to provide additional [Ti] and the consistency of [Ti] of different pits in the core of grain #8 precludes the result of non-Henry's law behavior in the incorporation of Ti in zircon. Therefore, the high-[Ti] grains (especially grain #8) suggest a crystallization condition different from the common Jack Hills zircons.

The δ18O value (7.2‰) of the core of grain #8 is similar to the highest values for the Jack Hills magmatic zircons (Fig. 6; Table S4) and is higher than the mantle-like magmatic zircon range (5.3 ± 0.6‰, 2SD)38. This implies that incorporation of altered crustal material into the magma source, or isotopic exchange of the protoliths with surface water occurred at 4150 Ma ago, which is consistent with previous conclusions4,9,15,38,39,40. The core of grain #8 may have been formed in a granitoid from the melting of supracrustal sediments as Jack Hills zircons and its high temperature from high-[Ti] indicates a water-free (dry) melting condition that has not been found from Jack Hills zircons.

Figure 6
figure 6

A compilation showing the distribution of 207Pb/206Pb age versus δ18O for the published analyses on >3300 Ma zircons and the magmatic areas of the two grains of this study.

The value for mantle-like zircon (5.3 ± 0.6‰, 2SD) is from the ref. 19. This plot is made on a compilation (Table S4) of 339 magmatic zircons. Only con% between 80% and 120% are included when multiple isotopic ages were provided. The analyses with δ18O errors (2SD) bigger than 1.8 were rejected. Detailed rules in data filtration can be seen in the notes of Supplementary Table S3.

Full size image

Oxygen fugacity calculations based on Ce/Ce* anomalies (see equation (2) in the supplementary files) have been used to constrain the oxidation state of early continental crust33,40,41. Note the calculations are mainly controlled by [Ti], [La], [Ce], [Pr] and Ti-in-zircon thermometer and thus the uncertainty of logfO2 is constrained by the uncertainties of these factors. The two studied grains show large positive Ce/Ce* anomalies (Table S5; Fig. 3). Two analyses (#TE8.1 and #TE8.5) of the grain #8 core give Ce/Ce* of 65 and 19, respectively, corresponding to logfO2 of −5.3 (ΔFMQ + 7.3) and −9.7 (ΔFMQ + 2.8) at 910°C (Table S2). The two logfO2 estimates overlap within uncertainty (Fig. 5) and their mean value (ΔFMQ + 5) is higher than most of published crust-derived Jack Hills zircons (Fig. 5). Although the oxygen fugacity estimates are based on many assumptions41,42, comparisons at same assumptions are still meaningful in particular for the rare Hadean zircons. The Ce/Ce* anomalies of the two core analyses of grain #8 are just within the range of published Hadean zircon data33 (Table S5), suggesting that the high T is the major cause for the high logfO2 according to the equations of Ti-in-zircon thermometry26,27 and oxygen fugacity41,42. Earth's earliest atmosphere had no significant free oxygen because of the absence of photosynthetic prokaryotic organisms, although the early mantle had similar oxygen fugacity as Archean and modern times40. The high-logfO2 melt for the grain #8 should generate from melting of crustal rocks which originally formed from locally mantle-derived high-logfO2 melt which has not been found in Jack Hills zircons yet. This suggests that the oxygen fugacity of early mantle may be heterogeneous.

The early earth crust may be heterogeneous as revealed by the variations in Hf, Li and O isotopes in Jack Hills Hadean zircons3,5,9,15,36,43, multiple age domains within single zircon grains44 and possible heavy bombardment epoch or early earth45. Based on the aforementioned discussion, the two 4100 Ma zircon grains (especially grain #8) from southern China are distinct from the published Jack Hills zircons in their high [Ti] and logfO2. Although a few published Jack Hills zircons showed complexity in CL texture, the pre-3.6 Ga multiple (possibly three episodes according to CL and U-Pb dating results) overgrowths of grain #8 and the 4070–4060 Ma metamorphic mantle are unusual in comparison to Jack Hills zircons. Two possibilities can be addressed for the origins of the two 4100 Ma zircon grains from southern China. Perhaps they were derived from a source area in Western Australia that differed from that of the Jack Hills zircons, or alternatively they could be from a different landmass never attached to the Yilgarn Craton of Western Australia. Further U-Pb dating on other detrital zircons of the studied sample and comparisons with the U-Pb age patterns of Western Australia will be useful way to distinguish them. If the second possibility is correct, Hadean landmasses may have been more widespread than previously known. Either way, the existence of high-[Ti]-δ18O-[Ce] zircon and the possible dry melting at ca. 4100 Ma and its 4070–4060 Ma metamorphism suggest diverse tectonic regimes of the Hadean crust.

It should be noted that all four known > 4050 Ma zircons dated by SIMS from China (including this work and the two xenocrysts from Tibet and North Qinling Belt20,21) are from Phanerozoic sedimentary/volcanic rocks, which is different from the Australia, Greenland and Canada localities where ancient zircons were found in Archean rocks. In addition, rare 3900–4020 Ma detrital zircons were also found in Neoproterozoic and Paleozoic sediments in southern China46,47 and in Late Devonian sediments in Hexi Corridor of northwestern China48 using laser ablation-ICP-MS method. The sporadic occurrence of Hadean to Early Archean zircons in young sediments in China shows that the Hadean zircons can survive multi-stage crustal recycling. Compared to the other reported >4050 Ma zircons outside Australia, finding two old grains out of 214 zircons of one sample represents a high proportion (nearly 1%) and suggests that more >4100 Ma zircons can be found. Additional U-Pb dating on the detrital zircons in the Longquan area of the Cathaysia Block is necessary to evaluate whether the unusual geochemistry of Hadean zircons in this study is representative and to help constrain the earliest crustal evolution.

Conclusions

Two ~4100 Ma detrital zircons were found in a Paleozoic quartzite from the Longquan area of southern China. One zircon shows normal magmatic oscillatory zonation in CL and constant mantle-like oxygen isotopes (δ18O = 5.8–6.0‰). The other zircon grain has a ~4100 Ma magmatic core surrounded by three >3600 Ma overgrowths, especially a 4070–4060 Ma metamorphic rim. The magmatic core is distinct in its elevated δ18O (7.2‰), high titanium concentration (53 ppm) and a calculated high oxygen fugacity (ΔFMQ + 5) and crystallization temperature (910°C), suggesting a granitoid-like source generated from dry remelting of partly oxidizing supracrustal sediments altered by surface waters. The unusual melting condition and the immediately following ~4070 Ma metamorphism recorded in zircon possibly provide new evidence for diversity of the Earth's earliest continental crust and more ancient zircons other than Western Australia as well as other isotopic and geochemical investigations are necessary to understand the earliest crustal evolution of the Earth.

Methods

After crushing and grinding, zircons were separated by heavy liquid and magnetic techniques. They were mounted in epoxy with the standard TEMORA 1 (206Pb/238U age = 417 Ma49) and polished to mid-section. Detailed cathodoluminescence (CL) images were made for each surface of analysis (Supplementary Fig. DR3). U-Th-Pb zircon analyses were performed on the SHRIMP II ion microprobe at the Beijing SHRIMP center, Chinese Academy of Geological Sciences, following standard operating techniques14,50. There are two dating sessions. Four U-Th-Pb isotope analyses (session-1; Table 1) were initially made for each grain (ion beam ~ 30 μm dia.) on surface-1 (Fig. DR3a–d).

After the session-1 dating, the mount was ground lightly and repolished to remove the SIMS pits. Further CL and SEM imaging (surface-2; Fig. DR3e–j) were performed for oxygen isotope and trace element analyses. Oxygen isotope ratios and trace element compositions were analyzed on surface-2 using the CAMECA IMS-1280 ion microprobe in the WiscSIMS Laboratory, UW-Madison, with detailed analytical conditions and data reduction procedures reported elsewhere51,52. Oxygen isotopes were analyzed (ion beam ~8 × 9 μm) with a zircon KIM-5 (δ18O = 5.09‰ VSMOW)53 as standard. A 133Cs+ primary ion beam (20 kV total impact voltage, 1.9–2.2 nA) was focused to an area of 8 × 9 μm on the sample surface. Total analytical time per spot was about 4 minutes: including pre-sputtering (10 s), automatic retuning of the secondary beam (120 s) and analysis (80 s). Trace element analyses were performed directly on the same pits as δ18O (beam size ~10 × 12 μm) at WiscSIMS in single collector mode by axial electron multiplier using magnetic peak switching. Zircon 91500 and NBS610 glass were used as standards following the method of Page et al.32 and Fu et al.54. A primary 16O− beam with a current of 2.5 nA and a total impact energy of 23 kV was defocused to a 10 × 12 μm spot. An energy offset of 40 eV was applied and the mass resolving power was set at 5000. All analyses consisted of 100 s pre-sputtering, 80 s for centering ions to the field aperture using the30Si+ signal followed by seven mass scan cycles for detection of trace element signals, with total analytical time about 24 minutes. Only the last 5 cycles were integrated; the first two cycles were used to stabilize the magnet. After analysis, each pit was imaged by SEM (Scanning electron microscope). No oxygen or trace element analyses have unusual pits and no cracks or inclusions were found in the pits.

The session 2 (Table 1) U-Pb dating analyses were made after oxygen isotope and trace element analysis at the Beijing SHRIMP center. Zircons were reground and repolished to remove all previous analytical pits; the grains were then re-imaged by CL (surface-3; Fig. DR3k–n). Care was taken to locate new analytical sites away from identifiable cracks and the primary ion beam size was adjusted to about 15–20 μm to avoid overlapping of different zones seen by CL. This proved partially successful, with four of six analyses obtained from each grain showing better than 89% concordance (Table 1).