ARTICLE DOI: 10.1038/s41467-018-07527-w OPEN Degassing-induced fractionation of multiple sulphur isotopes unveils post-Archaean recycled oceanic crust signal in hotspot lava Patrick Beaudry1,5, Marc-Antoine Longpré1,2, Rita Economos3, Boswell A. Wing4,6, Thi Hao Bui4 & John Stix4 Mantle source regions feeding hotspot volcanoes likely contain recycled subducted material. Anomalous sulphur (S) isotope signatures in hotspot lavas have tied ancient surface S to this 1234567890():,; deep geological cycle, but their potential modification by shallow magmatic processes has generally been overlooked. Here we present S isotope measurements in magmatic sulphides, silicate melt inclusions and matrix glasses from the recent eruption of a hotspot volcano at El Hierro, Canary Islands, which show that degassing induces strongly negative δ34S fractionation in both silicate and sulphide melts. Our results reflect the complex interplay among redox conditions, S speciation and degassing. The isotopic fractionation is mass dependent (Δ33S = 0‰), thus lacking evidence for the recycled Archaean crust signal recently identified at other hotspot volcanoes. However, the source has an enriched signature (δ34S~+ 3‰), which supports the presence of younger 34S-rich recycled oceanic material in the Canary Island mantle plume. 1 School of Earth and Environmental Sciences, Queens College, City University of New York, Queens, NY 11367, USA. 2 Earth and Environmental Sciences, The Graduate Center, City University of New York, New York, NY 10016, USA. 3 Department of Earth Sciences, Southern Methodist University, Dallas, TX 75275, USA. 4 Department of Earth and Planetary Sciences, McGill University, Montréal, QC H3A 0E8, Canada. 5Present address: Department of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA 02139, USA. 6Present address: Department of Geological Sciences, University of Colorado Boulder, Boulder, CO 80309, USA. Correspondence and requests for materials should be addressed to P.B. (email: [email protected]) NATURE COMMUNICATIONS | (2018) 9:5093 | DOI: 10.1038/s41467-018-07527-w | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-07527-w hemical, physical and biological processes within the spanning the degassing history of the volatile- and sulphur-rich mantle, crust, atmosphere and oceans fractionate sulphur magma erupted at El Hierro, Canary Islands, in 2011–201217.We C 34 Δ33 isotopes, and material exchanges among these geological also present δ S and S signatures of magmatic sulphide reservoirs lead to characteristic sulphur isotope signatures that inclusions — to our knowledge, this is the first time S isotopes are have varied over time1. Sulphur isotope heterogeneity in the measured simultaneously in sulphides, melt inclusions and matrix mantle, as sampled by sulphide inclusions2–5, melt inclusions6,7 glass for a single eruption. The dataset provides an unusually clear or primitive volcanic rocks8,9, thus traces secular variation in picture of the mechanisms by which sulphur isotopes fractionate the tectonic boundary conditions that influence mantle circula- during degassing and sulphide saturation in natural magmas, and tion, as well as the internal processes that have established shows that degassing can induce large δ34S fractionations of up to the current mantle state. On the other hand, the full scope of 10‰. In turn, this offers an exceptional opportunity to investigate mantle heterogeneity is best illustrated by variations in trace in situ how the S isotope heterogeneity generated during magma element and radiogenic isotope geochemistry of ocean island evolution, ascent and eruption7,8,18,19 can be quantitatively dis- basalts (OIB), which define various mantle end-members thought criminated from that inherited from the mantle sources for to feed the sources of hotspots10,11. Therefore, it is of interest to hotspot volcanism. In doing so, we find that the S isotope signals investigate these mantle components from a sulphur isotope at El Hierro reflect a post-Archaean origin for recycled S in the perspective to help set additional constraints on their origin and Canary Island hotspot, contrasting with the recent findings of S- further our understanding of the deep sulphur cycle. This type of MIF at other OIB localities. approach has been adopted in recent studies, which we sum- marise below. Results Mass-independent fractionation (MIF) of sulphur isotopes is Geological setting and sample description. The Canary Island defined by non-zero Δ33S values, where Δ33S = δ33S – [(1 + δ34S) hotspot in the eastern Atlantic Ocean is characterised by an exotic 0.515–1] and δxS = (xS/32S) /(xS/32S) – 1, and is V-CDT sample V-CDT geochemistry, producing OIB that is mostly alkaline in compo- characteristic of sedimentary rocks of Archaean age (~2.5 Ga and sition and displays isotopic affinities with the HIMU, EM and older), reflecting the influence of photochemical processes in an depleted MORB mantle (DMM) end-members20,21, suggesting atmosphere devoid of oxygen12. Sulphur in sedimentary rocks that the mantle source constitutes a mix of different reservoirs21. that post-date the Great Oxidation Event at ~2.4 Ga thus has Δ33S This context thus offers the potential to confirm or contrast S values around 0‰, and the associated S isotope fractionations isotope signals from the Canary Islands to those observed at are termed mass-dependent. However, negative, non-zero Δ33S hotspots in the South Pacific (i.e. Mangaia/HIMU; Pitcairn/EM-I; values have now been reported twice for young volcanic rocks Samoa/EM-II). Additionally, ubiquitous fluid inclusions in from hotspot settings: in olivine-hosted sulphides from Mangaia, mantle xenoliths and occurrence of carbonatite melt22,23 point to Cook Islands4, and in sulphides from the Pitcairn hotspot5. These a mantle source enriched in volatiles. The recent submarine anomalous signatures are thought to reflect the cycling of eruption off the south coast of El Hierro, the youngest and Archaean sulphur from Earth’s surface to the mantle by sub- westernmost island of the archipelago, produced lava balloons duction, and back to the surface via mantle plumes4,5,13. Since containing olivine-hosted melt inclusions (Fig. 1a) with dissolved Mangaia is the representative end-member of the ancient ‘HIMU’ volatiles reaching concentrations in excess of 3000 ppm CO ,3 (high μ = 238U/204Pb) mantle component10,11, the potential 2 wt.% H O and 5000 ppm S (ref. 17). In addition, clinopyroxene positive covariation of Δ33S and δ34S values in Mangaia sulphides 2 and spinel (Fe–Ti oxide) phenocrysts in the same samples host was first used to suggest a specific Archaean protolith to the abundant sulphide globules (Fig. 1b–f), revealing that the magma HIMU source characterised by negative Δ33S and δ34S values4. was saturated with an immiscible sulphide liquid for part of its The subsequent finding of S-MIF at Pitcairn, representative of history. However, sulphide inclusions are not present in olivine the enriched mantle I end-member10,11 (EM-I; characterised phenocrysts, nor do they occur as a free phase in the matrix glass. by unradiogenic Pb isotope signatures), in association with We performed in situ sulphur isotope analyses by Secondary negative δ34S values, lends support to this hypothesis, potentially Ion Mass Spectrometry (SIMS) on a suite of 25 olivine- and resolving the positively-skewed imbalance of Δ33S values spinel-hosted melt inclusions and 9 matrix glass chips, previously observed in Archaean surface reservoirs5,13. These studies thus shown to have large ranges in volatile contents and sulphur imply that a missing Archaean sulphur pool is stored in the speciation, with S contents linearly and positively correlated with deep mantle and occasionally resurfaces at hotspots. Other + H O and S6 /ΣS (ref. 17). These measurements yielded δ34S plume-related lavas from Samoa, the type locality for the third 2 values only, owing to the analytical difficulty of resolving low- common OIB mantle isotopic end-member, EM-II (characterised abundance 33S at low levels of S in silicate melts. We also by the highest 87Sr/86Sr ratios), show coupled variations in S obtained the S isotope compositions (δ34S and Δ33S) of 49 and Sr isotopes that indicate recycling of younger sulphur- clinopyroxene- and spinel-hosted sulphide droplets, and mea- rich sediments into a mantle source with a near-zero Δ33S sured their chemical composition by electron probe micro- value (mass-dependent) and a δ34S value of ~3‰9. The distinc- analysis (EPMA). Table 1 summarises the isotopic composition tion in S isotope signatures between different mantle reservoirs of the various samples. Details on our analytical techniques suggests a long-lived and isotopically evolving surficial input and associated uncertainties can be found in the Methods into different hotspot source regions, highlighting the importance section. The Supplementary Information includes a discussion of understanding the causes of S isotope variability in the mantle. of potential matrix effects during isotopic analysis, which appear Magmatic processes involving sulphur, such as degassing, negligible. All quoted errors are 1σ propagated analytical sulphide segregation, mixing or assimilation, may also leave an uncertainties (see Methods for error treatment). imprint on the S isotope composition of volcanic rocks7,8,14,15, and separating these effects from the source signature can present challenges. However, melt inclusions,