Report

Sulfur isotopes in diamonds reveal differences in continent construction

See allHide authors and affiliations

Science  26 Apr 2019:
Vol. 364, Issue 6438, pp. 383-385
DOI: 10.1126/science.aaw9548

Sulfur tells tectonic secrets

Plate tectonics dominates how the surface of Earth is shaped over geologic time; however, we do not know when this important process started on Earth. Smit et al. used sulfur isotopes in diamonds to show that atmospheric sulfur was entering the mantle 3 billion years ago through plate subduction. Older diamonds do not have this signature, constraining when plate tectonics started on Earth to around 3 billion years ago.

Science, this issue p. 383

Abstract

Neoproterozoic West African diamonds contain sulfide inclusions with mass-independently fractionated (MIF) sulfur isotopes that trace Archean surficial signatures into the mantle. Two episodes of subduction are recorded in these West African sulfide inclusions: thickening of the continental lithosphere through horizontal processes around 3 billion years ago and reworking and diamond growth around 650 million years ago. We find that the sulfur isotope record in worldwide diamond inclusions is consistent with changes in tectonic processes that formed the continental lithosphere in the Archean. Slave craton diamonds that formed 3.5 billion years ago do not contain any MIF sulfur. Younger diamonds from the Kaapvaal, Zimbabwe, and West African cratons do contain MIF sulfur, which suggests craton construction by advective thickening of mantle lithosphere through conventional subduction-style horizontal tectonics.

Earth’s oldest continents are stabilized by the presence of thick lithospheric mantle keels. These mantle keels have different physical and chemical properties than the upper mantle, which isolate them from mantle convection. Models for the formation of these mantle keels range from horizontal tectonic processes, such as advective thickening (1) or stacking of subducted oceanic lithosphere (2, 3) accompanying Wilson cycle continental collision (4), to vertical tectonic processes, such as subcretion of mantle plumes (5), melting at the base of oceanic plateaus (6), or diapiric ascent of foundered residues of plume melting (7). These divergent models have led to debate between proponents of horizontal versus vertical processes in the formation of the earliest continental lithosphere, because they provide important constraints on the geodynamics of the Archean Earth and the onset of plate tectonics. High-precision analyses of multiple sulfur isotopes provide a spatial and temporal geochemical tracer that can test for Archean surficial components in the continental lithosphere and reveal the construction style of continents at mantle depths.

Sulfur has four stable isotopes (32S, 33S, 34S, and 36S), whose abundances vary with mass during normal igneous and metamorphic processes in the solid Earth. Deviations from terrestrial mass-dependent isotopic compositions are known as mass-independent fractionation and produce anomalous 33S/32S and 36S/32S relative to 34S/32S (expressed as ∆33S and ∆36S, respectively) (8). Mass-independently fractionated (MIF) sulfur is evident in Hadean to Archean sedimentary sulfides, yet it mostly disappears from the sulfides of Proterozoic and younger sediments (9, 10). This change in sulfur chemistry is thought to be an atmospheric effect that was caused by the oxygenation of the atmosphere between 2.5 and 2.3 billion years (Ga) ago (9, 10). Before 2.5 Ga ago, mantle-derived sulfur was volcanically erupted into the stratosphere as SO2 and H2S [with 34S/32S (δ34S) values from −1.9 to +0.5 per mil (‰)] (11, 12), where it would have been subjected to photolysis by solar ultraviolet (UV) radiation, resulting in mass-independent isotope fractionation. After the Proterozoic oxygenation of the atmosphere, this effect disappears because ozone shields sulfur from UV interaction.

Because MIF sulfur isotopes are only generated by atmospheric processes in the Hadean-Archean, they are combined tracers of both age and exposure to Earth’s sedimentary cycle. This isotopic signature has emerged as a key way to trace the source of old, Archean sulfur in ocean island basalts (13, 14), crustal massive sulfide ores (15, 16), and sulfide inclusions in diamonds (17, 18). Although rare in samples derived from the convecting mantle (13, 14), such signatures can appear in ancient portions of some of the stable, 150- to 200-km-thick cratonic mantle roots that are resistant to convective mixing. We show that because multiple sulfur isotopes have the ability to trace the transport of Hadean-Archean atmospheric sulfur into the deep continental lithosphere, they add crucial geochemical evidence about the way the stabilizing keels below the continental lithosphere formed.

Zimmi alluvial diamonds contain abundant eclogitic sulfide inclusions and were derived from >120 km depths in the West African lithospheric mantle (19). The age of the Zimmi diamonds was determined using Re-Os isotopes, whose compositions fall along Neoproterozoic age arrays (19). We analyzed the sulfur isotopic compositions (δ33S, δ34S, and δ36S) of six sulfide inclusions in six different Neoproterozoic Zimmi diamonds using a Cameca IMS-1280 multi-collector secondary ion mass spectrometer (8).

The six sulfides are all pyrrhotite-pentlandite-chalcopyrite assemblages. Although sulfides show some slight internal variation in Cu and Ni content (19), we observed no correlation between major element composition and sulfur isotopic composition. Zimmi sulfides have δ34S values between −0.93 and +1.76‰ (Fig. 1 and table S1). These δ34S values broadly overlap with the field for depleted mantle and chondrites (−1.9 to +0.35‰) (11, 12, 2022). However, the sulfides in Zimmi diamonds all have low Ni contents (<7%) and high 187Os/188Os, supporting a recycled origin for the sulfur in the sulfides (19).

Fig. 1 33S and δ34S values for Zimmi sulfides.

(A) ∆33S and δ34S values for Zimmi sulfides relative to the experimentally produced arrays for reduction and oxidation of atmospheric SO2 (10, 13, 14). The δ34S range for volcanic SO2 is given as the full range reported for both depleted mantle and chondrite (−1.9 to +0.35‰) (11, 12, 2022). MIF sulfur isotope anomalies are thought to only arise in the Archean oxygen-poor atmosphere through photolysis of SO2 by UV radiation. However, none of the experimentally determined fractionation arrays (10, 41, 42) match the sulfur isotopic compositions of Archean metasediments (ARA). (B) Zimmi sulfide inclusion data compared with previous analyses of sulfide inclusions in diamond (17, 18, 23). Sulfide inclusions in Jwaneng and Orapa diamonds from the Kaapvaal and Zimbabwe cratons show resolvable MIF sulfur isotope anomalies, whereas sulfide inclusions in Paleoarchean Panda diamonds from the Slave craton do not. Error bars are given as 2σ uncertainties reported at 95% confidence level (8).

All six sulfide inclusions have δ33S and δ36S values that deviate from the arrays that define terrestrial mass-dependent fractionation, i.e., ∆33S and ∆36S = 0 (Figs. 1 and 2 and table S1). We divided the Zimmi sulfides into two isotopic groups. Group 1 has four samples with ∆33S between +0.36 and +0.75‰ and ∆36S between −0.36 and +0.42‰, and group 2 has two samples with ∆33S between +1.22 and +1.45‰ and ∆36S between −1.77 and −0.46‰. The average 2σ uncertainty on ∆33S is ±0.19‰, and ±0.77‰ on ∆36S (8).

Fig. 2 33S and ∆36S values for Zimmi sulfide inclusions.

33S and ∆36S values for (A) Zimmi sulfide inclusions analyzed in this study compared with (B) Archean metasediments from the literature (9, 24, 25). The slope of ∆36S-∆33S values in Zimmi sulfide inclusions is around −1, similar to the −0.9 slope of Archean metasediments (9) and experimental products after UV photolysis of SO2 (∆36S/∆33S: ~−1.1) (10). Sulfide inclusions in diamonds were previously only analyzed for ∆33S (9, 18, 23) (Fig. 1) and are given as a range in (B). Error bars are given as 2σ uncertainties reported at 95% confidence level (8).

Sulfides from group 1 have ∆33S values that overlap with the previously reported diamond sulfide compositions (∆33S = −0.5 to +0.91‰) from Jwaneng and Orapa on the Kaapvaal-Zimbabwe cratons (17, 18). Group 1 does not overlap with sulfides (∆33S = −0.35 to +0.23‰) from the Slave craton diamonds (23). Sulfides from group 2 have ∆33S > +1‰, which is far greater than previously reported for sulfide inclusions. In both groups, ∆33S-δ34S and ∆36S-∆33S relationships of the Zimmi sulfide inclusions overlap with the Archean reference array (ARA) for sulfur-bearing sediments (Figs. 1 and 2). In ∆36S -∆33S (Figs. 1 and 2), the slope we observed in Zimmi sulfide inclusions (−1.01 ± 0.12) is very similar to the −0.9 slope common in Archean metasediments (24, 25), suggesting that the sulfides incorporated atmospherically modified sedimentary sulfur.

The differences in sulfur isotope compositions between the two groups of Zimmi sulfides (Figs. 1 and 2) could reflect heterogeneity in sulfur at the time of Archean subduction and/or Neoproterozoic diamond formation. Heterogeneity at the time of Archean subduction could have been due to sulfides recording variable fractionation of atmospheric sulfur (variable ∆33S values) (Fig. 1). Alternatively, the Archean atmospheric-derived sulfur could have been mixed with some mantle sulfur (or other source of sulfur with mass-dependently fractionated isotope ratios) during primary sulfide formation in the Archean, during metasomatism over their 2-Ga mantle residency, or during Neoproterozoic formation of diamond. Mixing with mass-dependent sulfur would result in dilution of the primary MIF isotope signatures.

Regardless of the cause of the heterogeneous ∆33S and ∆36S values observed in Zimmi sulfides, the presence of MIF sulfur isotopes unambiguously confirms that sulfur had an Archean surficial prehistory and did not originate from Neoproterozoic oceanic crust, as might have been expected from the much younger 650-million-year (Ma) age of the diamonds (19). In diamonds of Neoproterozoic age, the presence of Archean sulfur in multiple sulfides from the same diamond, all with identical radiogenic 187Os/188Os and low time-averaged Re/Os, indicates that the Re-Os isotopic system was reequilibrated at the time of diamond formation. These features demonstrate that the Re-Os ages are robust, regardless of whether the sulfide inclusions have a prehistory or formed from the same fluid as the diamond (19).

Archean MIF sulfur isotopic compositions can be retained during metamorphism, subduction, and arc magmatism (26, 27). Data from modern Pacific ocean island basalts show that Archean sulfur isotope signatures can be preserved even at the high temperatures (>1300°C) in the convecting mantle (13, 14). Because MIF sulfur isotope anomalies have proven to be preserved in the convecting mantle, these anomalies can be expected to be even more robust in the cratonic lithosphere, which is cooler and has been isolated from convective mixing for billions of years.

A likely time for sulfur incorporation in the West African cratonic lithosphere was during subduction of oceanic lithosphere at 3 to 2.9 Ga ago, associated with emplacement of eclogites, formation of continental crust, and growth of the lithospheric mantle (28, 29). Mesoarchean crust formation was followed, more than 2 billion years later, by another episode of subduction beneath the existing craton during the Neoproterozoic (700 to 550 Ma ago) (30). Two separate subduction events documented in one suite of sulfide inclusions from the deep continental lithosphere demonstrate that subduction processes were essential to the growth and modification of the West African craton over a 2-Ga time scale.

The Slave craton has the oldest sulfides analyzed (in diamonds with 3.5 Ga Re-Os ages) (31), which contain no MIF sulfur isotopic compositions (23), suggesting that the peridotitic diamond host lithologies became part of the continental lithosphere in some way that excluded substantial surface material. This is consistent with models proposed for formation of the earliest Slave cratonic lithosphere, which would have incorporated only mantle sulfur, whether by vertical upwelling of depleted mantle residues (32), crust formation in an oceanic plateau setting (6), or by collisional compression (33, 34). All these modes of formation would have precluded incorporation of atmospherically modified sulfur. Only during later Meso-Neoarchean growth and stabilization of the Slave craton did lithosphere construction involve horizontal subduction-style tectonics that could incorporate surficial sulfur (35). In the sulfur isotope record, this model could be evaluated through analyses of sulfide inclusions from any <2.5-Ga-old diamond suites.

Atmospherically modified MIF sulfur incorporated into the continental lithospheric mantle, and preserved as sulfide inclusions in diamonds, emerges as a strong indicator of how continental lithosphere is formed and modified. Before the initiation of global subduction in the Archean, atmospherically modified MIF sulfur had no pathway for incorporation into the mantle. After subduction initiation, MIF sulfur could be expected in the mantle (Fig. 3). This signature is present in the younger diamonds from the Kaapvaal, Zimbabwe (17, 18), and West African cratons (this study) but not in the older diamonds from the Slave craton (23). Our results suggest that initial formation of the earliest Hadean to Paleoarchean proto-cratonic mantle roots did not involve recycling of surficial material [subduction stacking (2)]. Rather, horizontal accretion processes only became dominant during later Meso-Neoarchean accretion, causing thickening and stabilization of the Kaapvaal and West African continental lithospheres (28, 29, 36).

Fig. 3 Sulfur isotopes as a tool to understand the geodynamics of cratonic mantle construction.

MIF sulfur has both a distinctive Hadean-Archean time stamp and an atmospheric link that emerges as a very sensitive way to track Archean surficial sulfur into the deep mantle below continents. (A) Sulfides in Paleoarchean diamonds from the Slave craton do not contain MIF sulfur, supporting models for craton construction that did not involve incorporation of recycled surficial material. (B) Younger diamonds from the West African, Kaapvaal, and Zimbabwe cratons contain MIF sulfur, which suggests construction of the cratonic mantle through subduction-style horizontal processes.

Studies from other cratons show evidence for accretionary collisional processes in Neoarchean lithosphere (Yilgarn and Superior cratons) (37, 38) but the absence of collisional processes before 3.2 Ga ago (Pilbara craton) (39). These differences represent two distinct styles for the construction of continental mantle roots and are consistent with a Mesoarchean transition in mantle tectonics to predominantly horizontal processes, when plate tectonics became crucial for the growth, thickening, and stabilization of continents (4, 35, 40).

Because of the robustness of MIF sulfur isotopes in the cratonic lithosphere, analyzing sulfur isotopes in diamonds with sulfide inclusions is a sensitive way to trace the origin of mantle-derived sulfur. Unlike other isotopic analyses of elements like carbon, nitrogen, oxygen, rhenium, and osmium in diamond inclusions, MIF sulfur also has a distinctive Hadean-Archean time stamp. Sulfur isotopes in diamonds and their inclusions can be combined with Re-Os ages to track multiple subduction events during craton growth, even those separated by billions of years. This combined approach provides a powerful method for investigating the construction of continental mantle keels.

Supplementary Materials

science.sciencemag.org/content/364/6438/383/suppl/DC1

Materials and Methods

Tables S1 and S2

References (4348)

References and Notes

  1. Materials and methods are available as supplementary materials.
Acknowledgments: We thank I. Eliezri (Coldiam Ltd.) for providing samples from Zimmi, A. Chan (GIA) for his assistance during the laser cutting and polishing of diamonds, and R. Dokken (Alberta) for assistance with SIMS mount preparation. Our coauthor Erik Hauri will be sorely missed. We thank anonymous reviewers for their constructive reviews that improved the paper. Funding: This work was supported by GIA, University of Alberta, NSF grant EAR-1049992 (S.B.S. and E.H.H.), and the Carnegie Institution for Science, and it is a contribution to the Deep Carbon Observatory. Author contributions: K.V.S. led the research, polished and characterized the samples, interpreted results, and wrote the initial manuscript. S.B.S. contributed to scientific interpretations and manuscript writing. E.H.H. conducted ion probe analyses and contributed to scientific interpretations and the writing of the initial manuscript. R.A.S. conducted ion probe analyses. Competing interests: The authors declare no competing interests. Data and materials availability: All data for samples and reference materials are available in the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article