One of the most accurate and complete records for tracing the weathering of continental crust over geological time are strontium isotopes (87Sr/86Sr) in marine chemical sediments. Strontium in seawater is derived from two sources with distinct radiogenic isotopic compositions: hydrothermal alteration of oceanic crust with low and mantle-like 87Sr/86Sr, and subaerial weathering of continental crust with higher 87Sr/86Sr 1. Because Sr has a long residence time relative to the ocean mixing rate, 87Sr/86Sr in seawater is globally homogenized and the balance between oceanic and continental inputs can be recorded in the Sr isotope ratios of authigenic marine minerals. Typically the lowest 87Sr/86Sr (least radiogenic) are taken as the best estimate of seawater at any time since post-depositional alteration is most likely to increase Sr isotope ratios 2. Throughout the Phanerozoic, carbonate shells and limestones show many secular variations in seawater 87Sr/86Sr that can be linked to changes in seafloor spreading rates, as well as shifting tectonic, geographic and climatic controls on the Sr isotopic composition of riverine runoff 3. Further back in time, the seawater Sr isotope record becomes compromised by the scarcity of unaltered sedimentary carbonate rocks. Sparse Archean carbonates with highly unradiogenic 87Sr/86Sr have therefore been interpreted to represent seawater, suggesting a mantle-dominated ocean chemistry at this time 4, 5, 6. This curve and the concomitant rise in seawater 87Sr/86Sr at the Archean-Proterozoic boundary supports models and proxies that argue for the late emergence of continental crust and onset of weathering around 2.5 Ga 7, 8, 9, 10, 11, 12, 13. However, recent work on two barite deposits suggested much higher seawater 87Sr/86Sr than the mantle-like value assumed from the Archean carbonate record at 3.2 Ga 14, 15. This finding questions the assumption that the unradiogenic carbonates truly reflect seawater, and challenges the validity of a mantle-like Sr isotope seawater evolution curve for the Paleoarchean. The earlier onset of weathering inferred from these barite data would be consistent with Ti isotopic evidence for emerged felsic crust at 3.5 Ga 16, weathering-induced decoupling of Hf and Nd isotope systems in 3.4 Ga chert 17, and Hf isotope ratios in black shales indicating weathering of evolved crust by 3.0 Ga 18.
To correctly reconstruct the onset of crustal weathering and emergence of continental crust, here we investigate the Sr, S and O isotopic compositions of six stratiform marine-hydrothermal barite deposits from three different cratons covering a time window of 320 million years. Using a mixing model to assess hydrothermal influence on the barite 87Sr/86Sr, we calculate a new Sr isotope evolution trend for Paleoarchean seawater and use this result to constrain the timing at which weathering began to modify early ocean chemistry.
Does barite reflect seawater?
We studied field and drill core samples from barite deposits at Londozi (3.52 Ga, Theespruit Fm), Vergelegen (3.41 Ga, Kromberg Fm), Stentor/Amo (3.26 Ga, Bien Venue Fm) and Barite Valley (3.24 Ga, Middle Mapepe Fm) in the Kaapvaal craton, North Pole (3.49 Ga, Dresser Fm) in the Pilbara craton and Sargur (3.20 Ga, Sargur Group) in the Dharwar craton (Table S1). All barite deposits occur in volcanic-sedimentary successions consistent with shallow to deep marine environments 19, 20, 21, 22, 23, 24, 25. Multiple sulfur isotope data provide evidence for atmospherically-derived sulfate in the barites from a well-mixed global seawater reservoir 26, 27, 28 or local felsic volcanic eruptions 29, 30. In contrast, field observations suggest a hydrothermal origin for the barium 25. In all localities, barite is strongly associated with chert 21, silica dykes feed into barite horizons at Barite Valley and North Pole 25, 31, and barium-rich hydrothermal alteration zones with Ba-feldspar underlie barite at Londozi and Sargur 19, 29. However, all deposits lack polymetallic sulfide deposits, indicating that hydrothermal fluid temperatures were relatively low and likely below 150°C 32. Low-temperature hydrothermal activity is also consistent with the lack of an underlying magmatic system, absence of sintering and the small vertical extent of chert dykes at Barite Valley25.
Two types of barite are observed in the six deposits: bladed barite consisting of course blades up to several centimeters long, and fine-grained granular barite (Fig. S1). Crystal morphology alone does not reveal the origin of the barite. However, a combination of sedimentological evidence and morphology supports a primary origin for bladed barite at Barite Valley, except for some isolated blades cutting through barite sands that appear to have grown diagenetically 23, 25. A primary origin for bladed barite is consistent with the well-formed, tabular to bladed crystal morphology that is predicted to grow in settings with low to moderate degrees of barite oversaturation 33 and is found in modern hydrothermal settings 34. In contrast, granular barite is often found in association with heavy minerals and reworked quartz, indicating a detrital origin for this morphological type 23, 25. Equigranular textures can also form by recrystallization of barite during diagenesis or metamorphism that has affected all deposits 35, from lower greenschist facies (300-400°C) at North Pole, Barite Valley and Vergelegen 36, 37 to upper greenschist facies (400-500°C) at Stentor/Amo 20 and amphibolite facies (500-650°C) at Londozi and Sargur 38, 39. In order to determine which barite can be used to constrain Paleoarchean seawater 87Sr/86Sr, field data and mineral morphology must therefore be integrated with geochemical proxies.
Within individual deposits, bladed barite samples are Sr isotopically distinct from granular barite. Measured 87Sr/86Sr values are lowest in bladed barite (Fig. 1a), with weighted averages (± 95% confidence intervals) ranging from 0.700562 ± 0.00015 at Londozi (n = 2), 0.700841 ± 0.00004 at Vergelegen (n = 6), 0.701295 ± 0.00008 at Barite Valley (n = 4) to 0.701333 at Sargur (Table S2). No bladed barite was found in samples from the Stentor/Amo deposit. In contrast, granular barite is characterized by higher average 87Sr/86Sr values of 0.700757 ± 0.00010 at Londozi (n = 21), 0.701112 ± 0.00001 at Vergelegen (n = 2), 0.701240 ± 0.00013 at Stentor/Amo (n = 5), 0.701478 at Barite Valley and 0.701814 ± 0.00033 at Sargur (n = 3). These higher values cannot be explained by in situ 87Rb decay, because measured Rb concentrations in acid-leached fractions from both types of barite are very low and would require corrections less than our analytical precision (Table S3). In addition, we carefully selected least weathered samples to avoid contamination with high 87Sr/86Sr phases, as exemplified by the two highly weathered samples from Barite Valley (Fig. 1a).
Unlike the marine carbonate record, the lowest 87Sr/86Sr values in bladed barite cannot be unambiguously interpreted to reflect seawater as ratios may have been lowered by hydrothermal input of unradiogenic Sr 15. We therefore combine 87Sr/86Sr data with oxygen and sulfur isotopic compositions to select which barite is most representative of seawater (Fig. 1b-d) 27, 40. Bladed and granular barite is characterized by δ18O and δ34S values that fall within the range reported for Paleoarchean seawater sulfate from sulfate minerals and carbonate-associated sulfate (Fig. 1b, 1c) 15, 26, 41, 42. Seawater-like δ18O and δ34S values do not give direct evidence for seawater Sr isotope ratios in hydrothermal barite, since the source of Sr is not directly coupled to that of SO42- 35. However, the observation above is consistent with the low-temperature hydrothermal settings inferred from field data, as modification of seawater-like δ18O values is expected above 150°C due to rapid oxygen isotope exchange between dissolved sulfate and water 43. This in turn suggests that measured 87Sr/86Sr have been relatively little affected by Sr from a non-seawater source, because leaching of Sr from rocks is limited at these low temperatures and low- to intermediate-temperature hydrothermal fluids are dominated by seawater-derived Sr 44, 45.
Importantly, we observe the highest δ18O values for each deposit in bladed barite and in association with the most negative, and therefore most seawater-like 46, anomalous sulfur isotope signatures (D33S, see Methods for calculation, Table S2). These samples also display a strong positive correlation (R2 = 0.95) between 87Sr/86Sr and Δ33S (Fig. 1d), in contrast to a weaker correlation for granular barite (R2 = 0.64, not shown in Fig. 1d). Previous work has demonstrated that the magnitude of seawater sulfate D33S decreases throughout the Paleoarchean, as shown in Fig. 1d 26, 27, 42. The observed correlation between Sr and S isotopes in bladed barite is therefore best explained by co-evolution of D33S and 87Sr/86Sr in seawater due to progressive decay of 87Rb. In contrast, this correlation may have been blurred in the granular barite as a result of alteration or contamination, which is consistent with granular textures resulting from recrystallization processes. At water-rock ratios of 1 to 10, metamorphic fluids with 50-1000 ppm Sr and 87Sr/86Sr ~ 0.703-0.706 can shift Sr isotopic compositions from those measured in the bladed barite towards the higher values in granular barite samples (Fig. S2).
Based on the O, S and Sr isotope systematics outlined above, we conclude that 87Sr/86Sr of bladed barite is as close to Paleoarchean seawater as possible for a hydrothermal deposit. The 87Sr/86Sr values in the bladed barite samples define a strong regression line (Fig. 2, R2 = 0.98), and are more radiogenic than the Paleoarchean primitive mantle (Fig. 1a) calculated from the Basaltic Achondrite Best Initial (BABI) at 4.56 Ga of 87Sr/86Sr = 0.69897 47 and bulk Earth 87Rb/86Sr = 0.085 48. Values also plot above estimates of the depleted mantle based on 87Sr/86Sr = 0.703 for modern mid-ocean ridge basalt 49 and an initial value of 87Sr/86Sr = 0.69950 (DM1) 9 or 0.69897 (DM2) 47 (Fig. 1a). Our findings are consistent with previously reported Sr isotopic compositions of the Barite Valley, Sargur, Vergelegen and North Pole deposits 14, 15, 50. To go further, we explore next the hydrothermal influence on barite 87Sr/86Sr values to assess the Sr isotopic composition of the Paleoarchean oceans.
Paleoarchean seawater evolution trend
We use a hydrothermal mixing model to calculate a plausible seawater Sr evolution trend from the 87Sr/86Sr values measured in bladed barite samples, based on the low-temperature hydrothermal setting inferred from field and oxygen isotope data. In our model (see Methods), we calculate mixing ratios of seawater (20°C, 0.6M NaCl salinity) 51, 52 and hydrothermal fluid (150°C) that lead to oversaturation with respect to barite 53. We use a local seawater sulfate concentration of 8 µM based on observed 12-20‰ differences between barite and pyrite δ34S 54, 55, and assume the same SO42- concentration for the hydrothermal fluid as anhydrite precipitation and thermochemical sulfate reduction were likely negligible at 150°C. Our calculations for a range of Ba concentrations show that the highest degree of oversaturation occurs for a mixture consisting of 10-40% hydrothermal fluid (Table S5). Calculated saturation indices are low (<0.2), which is consistent with our interpretation above that bladed barite formed as primary crystals 33.
We use this result and a two-component mixing model 34 to calculate 87Sr/86Sr values of seawater-hydrothermal fluid mixtures from which barite precipitated. By varying the input value for seawater 87Sr/86Sr, we assess which seawater composition is feasible with the highest and lowest 87Sr/86Sr values measured in bladed barite from Londozi, Vergelegen and Barite Valley. To constrain the Sr isotopic composition of the hydrothermal fluids, we assume that Sr is predominantly derived from seawater (80%) with a small contribution from leached crust (20%), as observed in low-temperature (150°C) hydrothermal experiments 45 and modern low-intermediate temperature hydrothermal fluids 44, and consistent with seawater-dominated REE patterns in Paleoarchean alteration zones 32. For each deposit, we explore 87Sr/86Sr variations during hydrothermal leaching of mafic and felsic rocks, represented by Sr isotope evolution curves for the depleted mantle and continental crust, respectively.
From our constraints on the chemical conditions required for barite precipitation and comparison of theoretical seawater-hydrothermal mixtures with those recorded in the bladed barite, we obtain a seawater Sr isotope evolution trend with a slope corresponding to 87Rb/86Sr values of 0.194-0.198 (Fig. 2). Sr isotopic compositions of single bladed barite samples from North Pole and Sargur fall well within the predicted trend, as well as granular barite from the Stentor deposit. Calculated 87Rb/86Sr values are substantially higher than the estimated value for the early Archean depleted mantle (0.07 ± 0.007) 50 and contrast strongly with the mantle-dominated curve inferred from carbonate 87Sr/86Sr 4, 5, 6 (Fig. 3). Our results empirically constrain the seawater Sr isotope evolution trend significantly further back in time compared to the curve predicted from the extrapolation of 3.2 Ga barite 87Sr/86Sr 15, 56 (Fig. 3).
Onset of subaerial crustal weathering
The radiogenic Sr isotope values for 3.52-3.20 Ga Paleoarchean seawater imply detectable weathering of an emerged and felsic crustal source at least 300 million years further back in time than what has previously been reported at 3.2 Ga 15. We further advance this by constraining the start of subaerial weathering from the intersection of our seawater evolution trend and mantle curves, which represents the time at which the input of crustal Sr started to modify the 87Sr/86Sr of seawater away from mantle-dominated values. Our calculated trend for Paleoarchean seawater 87Sr/86Sr indicates an onset at approximately 3.7 ± 0.1 Ga based on the intersection with the primitive mantle curve, 3.6 ± 0.1 Ga from depleted mantle curve DM1 and 3.8 ± 0.1 Ga from depleted mantle curve DM2 (Fig. 2). Improved constraints on the Paleoarchean Sr isotope mantle curve are required to further reduce the uncertainty on this estimate. However, the late Eoarchean onset of crustal weathering observed from chemical signatures is consistent with the siliciclastic rock record which shows that physical weathering products appear for the first time in greenstone belts around 3.4 Ga 57 and possibly as early as 3.7 Ga 58 (Fig. 3), as well as examples of exposed land surfaces by 3.5 Ga 59 and evidence for the existence of felsic crust at this time 60. Our findings indicate that weathering substantially modified the Sr isotope budget of Paleoarchean seawater, in contrast to its oxygen isotope composition that was recently shown to be unaffected by this 12.
Comparison of our Sr isotope data and revised seawater trend with other records of weathering (e.g. 6, 11, 61) highlights the scarcity of data in the Paleoarchean and the importance of the barite record for this period in Earth’s history. Paleoarchean carbonate from the Pilbara and Kaapvaal craton 1, 5 displays significantly more radiogenic 87Sr/86Sr than barite (Fig. 3), reflecting the higher preservation potential of insoluble barite during diagenesis and metamorphism 35. The combined updated barite (blue lines in Fig. 3) and carbonate 56 (green line in Fig. 3) 87Sr/86Sr record for the Archean shows little secular variation from 3.7 to 2.5 Ga, which differs from the recently predicted sharp rise in seawater 87Sr/86Sr at 3.2 Ga 56. The absence of strong secular variations in the Archean 87Sr/86Sr record may be partially explained by the low temporal resolution of the data compared to the Phanerozoic, although a linear Sr isotope trend in the Paleoarchean is consistent with small 87Sr/86Sr variations in newly formed crust due to little time for radiogenic ingrowth of 87Sr. It also tentatively suggests no measurable long-term shifts in tectonic, geographic and climatic controls on seawater 87Sr/86Sr throughout the Archean, in contrast to the changes observed in carbonate 87Sr/86Sr from the Proterozoic and Mesozoic into the Cenozoic 62.
If the extrapolation of our seawater Sr isotope trend is correct, it implies that the late Eoarchean geodynamic regime generated granitic magmas and sufficient continental freeboard to support weathering of emerged felsic crust from 3.7 Ga. The globally significant changes in seawater 87Sr/86Sr defined by our samples from three different cratons suggest that subaerial weathering was a relatively widespread phenomenon, likely requiring a larger relative area of emerged crust than the 4% previously reported for the Neoarchean (see Supplementary Information) 63. Following the same reasoning as above, the absence of radiogenic Sr in seawater before 3.7 Ga suggests that prior to this time the extent of felsic landmass and subaerial weathering was very limited. This in turn would have hampered the colonization of land, inhibited the supply of nutrients to the oceans 64 and restricted the extent of epicontinental seas in the early Eoarchean, possibly limiting the evolution of photosynthetically fueled ecosystems. The lack of undisputed evidence for microbial activity in supracrustal rocks older than 3.5 Ga may reflect such an Eoarchean planet with fewer favorable environments for life to flourish than in the Paleoarchean, when crustal emergence and weathering facilitated life in shallow marine settings.