Introduction

Serpentinisation occurs when ultramafic rocks react with water. During serpentinisation, anhydrous olivine and pyroxene break down to form hydrous Fe2+-bearing minerals, Fe3+-bearing minerals, such as magnetite, and Fe0-bearing alloys such as awaruite1,2. Oxidation of Fe2+ leads to H2 production and highly reducing conditions2,3,4. The H2 produced by serpentinisation can reduce carbon dioxide (CO2) to form methane (CH4) and simple organic compounds4,5,6,7.

Nitrogen is also crucial for life and constitutes 10% of a cell by dry weight7. Arguments around the availability and formation of N-bearing species, particularly cyanide (CN), have been used to differentiate between origin-of-life theories involving (i) formation of the key building blocks of life from HCN produced by vaporisation of impactors8,9,10; and (ii) the ultra-reducing conditions associated with serpentinisation3,7 where H2-rich, alkaline fluids and Fe- and Ni-catalysts facilitate reduction of CO2 by H2 to form acetate, formate, and pyruvate. These reactions mimic, abiotically, the ancient acetyl-CoA metabolic pathway used by the last universal common ancestor (LUCA)7,11.

While serpentinites provide the necessary catalysts and reducing conditions to support the origin of life, N-sources and cycles in serpentinising systems are poorly known7,12. Peridotite protoliths of serpentinites are C- and N-poor, so C and N must be derived from the atmosphere7 or other sources9,13. Fluid inclusions in pyrrhotite associated with serpentine and magnetite from the asteroid Ryugu14and dust from the asteroid Bennu15host C- and N-species, indicating that these species can form abiotically. However, N sources, speciation, and cycling in terrestrial and extra-terrestrial serpentinising environments are cryptic and seldom observed16,17.

Catalysts are essential in metabolism, and catalysts such as awaruite (Ni3Fe) are fundamental to theories of the origin-of-life in serpentinising environments. Native copper (hereafter referred to as copper) acts as a catalyst in a range of natural18 and engineered19,20 systems. Alloys such as awaruite, and copper are common in mafic to ultramafic bodies in orogenic massifs and ophiolites21,22,23. Hence, relationships between known catalysts and C- and N-species in natural systems may provide insights into C- and N-cycling in serpentinites.

However, native metal grains are often only a few µm, or less, in size, and this makes them difficult to analyse. Bulk rock techniques such as inductively coupled plasma–mass spectrometry (ICP–MS) lack the spatial resolution to analyse them. In-situ techniques such as ICP–MS coupled with laser ablation (LA–ICP–MS), have a lateral resolution > 5 μm, and cannot detect H24. Electron microprobe analysis (EMPA) can analyse particles as small as ~2 µm25 but is limited to elements with atomic number Z > 526. In contrast, the combination of secondary ion mass spectrometry (SIMS) with a time of flight (ToF) detector gives a lateral resolution of ≤ 50 nm27and it is sensitive to elements with Z as low as 128. ToF–SIMS is a technique that produces isotopic and elemental maps of the sample surface using secondary ionised particles produced by surface bombardment with a primary ion (e.g., Bi3+). ToF-SIMS can map organic3,29 and inorganic30,31 components in geological materials at the nanoscale with high mass sensitivity (m/Δm of 5000–1000029). Atom probe tomography (APT) is a nanoscale technique that provides 3D, quantitative characterisation of mineral major, trace and isotopic composition, with spatial resolution as low as 1 nm32that has been used to analyse natural alloys21,33.

Here we document the micro- to nanometre-scale distribution of Cu, C, and N in serpentinising environments, using ToF–SIMS and APT on sulphides and copper grains within two serpentinised peridotite samples, BA3A-86-1 and BA4A-79-2, from the Wadi Tayin ophiolite (WTO), which is part of the Oman Ophiolite (Fig. 1). Details of the geological setting are provided in Supplementary Table 1. We use the results to: (1) understand the role of catalysts in C and N reduction; (2) elucidate the role of serpentinisation in the present-day crust–mantle N cycle; and (3) derive implications for the origin of life.

Fig. 1
figure 1

Geological map of the Oman Ophiolite and the Wadi Tayin Ophiolite (black box), part of the Southern Massif, in violet (modified after51).

Full size image

Results

Petrology

The mineral assemblage of samples BA3A-86-1 and BA4A-79-2 is typical of highly-serpentinised peridotites. The serpentine mineral lizardite4,34,35 replaces Mg-rich olivine and pyroxenes while magnetite partially replaces pentlandite (Fig. 2). Ragged chalcopyrite hosts copper grains (≤ 20 μm in size) and nanometre-scale pores (Figs. 2 and 3a). Cu is hosted by residual chalcopyrite and copper (Fig. 2, Supplementary Fig. 2).

Fig. 2
figure 2

(left) Backscattered Electron (BSE) image of the area of work (aow) 1 (a,b), 2 (c,d) and 3 (e,f). Images (b,d,e) are zoom-ins of the blue rectangle in (a,c,e), respectively. Magnetite (Mt) replaces pentlandite (Pn), forming veins and mesh textures. The chalcopyrite (Cpy) and secondary Mt embeds the copper grains (Cu, orange open circle and arrow); (right) Time of flight-secondary ion mass spectroscopy (ToF-SIMS) ion maps of the aow 1 (A,B), 2 (C,D) and 3 (E,F). Cu– is shown in (A,C,E), while CN– is in (B,D,F).

Full size image
Fig. 3
figure 3

(a) Backscattered electron (BSE) image of the pentlandite (Pn), partially replaced by magnetite (Mt), hosting the native copper grain investigated with the APT in the blue box; (b) zoom in of the copper grain and location of the APT specimen M19 (detail in c); (d) Reconstructed atom probe data showing Cu (orange) and Ag (light blue) nanoscale distribution in specimens M10 and M19. Cu atoms are shown at 4% (M10) and 7.29% (M19) of total Cu atoms, and the grey isoconcentration surface corresponds to 0.25% (M10) and 0.4% (M19) of Ag. The orange arrows in b and c point to the nanometer scale porosity.

Full size image

ToF–SIMS ion maps

The ToF–SIMS maps reveal that C- and N-species (CSN–, CS2N–, CNO–, aow, Fig. 1, Supplementary Fig. 1) are co-located with chalcopyrite and copper, with greater count rates for these species on chalcopyrite than on copper (Fig. 2A–F, Supplementary Tables 3, 4, 5). Complex charged molecular species, such as CSN–, may represent molecular fragments produced during the ToF–SIMS analyses and may not reflect the stoichiometry of species within the sample prior to analysis. However, the high spatial resolution of the technique (50 nm) indicates unambiguously that C, N, and S were co-located within the sample.

Possible sources of contamination include atmospheric C and N, hydrocarbon absorption on the specimen surface within the ToF–SIMS chamber, and the epoxy used for sample preparation, which contains C and N. To eliminate contamination of the signal by surface hydrocarbons, the areas for analysis were pre-sputtered with Cs+ to remove contaminants (Supplementary Table 2). The possibility of atmospheric contamination was reduced by equilibrating the sample for 30 min under ultra-low vacuum conditions (~ 10−9 mbar) prior to analysis. To check for an epoxy signal, areas of porous serpentine were also analysed. If epoxy contamination of the surface were present then it would be stronger within the porous serpentine than on the relatively non-porous chalcopyrite and native copper, but no signal for C or N was observed from the serpentine analyses (Fig. 2).

APT results

The APT analyses of two copper specimens show that Cu is the most abundant element (99 at% in M10 and 97.92 at% in M19) and that copper hosts nanometre-scale pores (Fig. 3b, c, Supplementary Fig. 2, 3). Neither C nor N were detected (Fig. 3d, Supplementary Table 6). The main impurity measured by APT in specimens M10 and M19 is Ag (0.004 at% and 0.027 at%, respectively), which is concentrated on planes (Fig. 3d).

Discussion

Based on these textural and geochemical observations, we propose a conceptual model that links C- and N-cycling during serpentinisation to chalcopyrite alteration (Fig. 4). Generally, olivine conversion to serpentine (Srp) can be described by Ol + H2O → Srp + magnetite (Mt) + H2 (Fig. 4, Eq. 1). Peridotitic protoliths are nominally free of N (< 1 ppm36) and C (< 100 ppm37), and are excluded as the source of these elements. The most likely source of C and N is dissolved CO2 and NO3– in serpentinising fluids, probably seawater or meteoric water38,39 (Fig. 4). Reduction of CO2 and NO3– to form C- and N-species is effected by electron transfer from H2, which is stabilised by the reducing conditions associated with serpentinisation and may be abiotic or biotically facilitated (Fig. 4, Eq. 3). The CO2 and NO3– reduction could be described by the reaction 5H2 + CO2 + NO3– → CN– + 5H2O (Fig. 4, Eq. 3). Reducing conditions also trigger desulphidation of chalcopyrite and reduction of Cu2+ to form secondary copper (Cu0, Fig. 4); reduction of CO2 and NO3– is catalysed by Cu18,19. These processes localise CO2 and NO3– to form C- and N-aggregates on chalcopyrite (Fig. 4, Eqs. 2, 3). Chalcopyrite (Cpy) reduction can follow the reaction 3Cpy + 2H2 + 4H2O → 3Cu0 + 6H2S + Mt (Fig. 4, Eq. 2). Evidence supporting the individual components of the model is discussed below.

Fig. 4
figure 4

Conceptual model showing the process that modified the WTO: serpentine (Srp) replaces olivine (Ol) and produces magnetite (Mt) and H (Eq. 1), which participates in the reduction of Cu2+ liberated during the chalcopyrite (Cpy) dissolution (Eq. 2). The microbiome may have participated in fixing Cu0 in the pores. The remaining H2 is consumed by microbiomes as for the NO3– introduced with the serpentinising fluids, leading to the precipitation of the CN– species (Eq. 3). *The stoichiometry of reaction [1] depends on the Mg# of Ol and Srp (Mg# = (MgO % / (MgO % + Fe2O3%)) × 100), and the extent to which aqueous Mg- and Si-bearing species are formed.

Full size image

Whole-rock analyses of peridotite samples from Oman indicate that C and N contents are elevated relative to mantle values, consistent with addition of C and N after crystallisation. Harzburgite samples from hole BA3A show whole-rock C concentrations of 215 ± 15 ppm to 1242 ± 86 ppm39considerably higher than the likely C content of the peridotite protolith. Harzburgite and dunite samples from nearby hole BA1B are also enriched in C, with concentrations of 467 ± 94 ppm to 1300 ± 40 ppm C in harzburgite and 296 ± 32 ppm to 13,608 ± 219 ppm in dunite39. The Oman DP rocks are also enriched in N relative to their likely peridotite protolith; whole-rock samples collected at 280 m depth in holes BA3A and BA4A have N concentrations of 13.9 ppm and 11.3 ppm, respectively40.

Carbon isotope analyses of carbonate veins within a harzburgite that crops out ~ 10 km from the Oman DP Holes indicate that C within the veins is a mixture of C derived from C-depleted mantle (δ13C < − 11‰), seawater (δ13C ~ 0‰), and meteoric water sources (δ13C = − 7.3–−8.3‰)38. Early veins show variable δ13C values of − 4.6‰ to − 15.2‰, later veins have δ13C values of − 4.1 to − 9.3‰38, and the latest veins have a δ13C signature similar to that of meteoric water (δ13C values of − 6.69 to − 8.3‰38). It is thought that the earliest veins formed before ophiolite emplacement, while type 2 and 3 veins formed after exhumation of the WTO in subaerial conditions38. Thus whole-rock C and N contents, and C isotope values, are consistent with addition of C from diverse sources pre- and post exhumation, and with addition of N post-crystallisation.

Chalcopyrite alteration and C- and N-reduction may proceed biotically or abiotically. There is little direct evidence of biotic activity within the studied WTO samples, but abundant evidence for bacterial activity within the WTO4,6,40,41,42,43 and circumstantial evidence supports biotic activity within at the site of chalcopyrite alteration and C- and N-compound formation.

For example, bacterial activity has been reported throughout the WTO. Laminated C- and N-structures (Frutexites) in the top 15 m of Hole BA1B are thought to be of microbial origin6. The Frutexites are overlain by a biofilm, partly constituted by Nitrosomonas, a chemoautotrophic ammonia-oxidising bacterium6. In addition, nitrogen-reducing bacteria are recorded by the presence of the narG, napA and nrfA genes in the metagenome of biomass dissolved in hyperalkaline waters sampled at depths of up to 400 m from OmanDP Hole BA3A and BA4A, and from 12 boreholes drilled by the Oman Ministry of Regional Municipalities and Water Resources, OMRMWR, which are within 5 km of OmanDP Hole BA3A and BA4A40,41. Previous work detected, from the 16Sr RNA genes of the waters in the wells, Thermodesulfovibrio (OmanDP4 and OMRMWR42 Holes), sulphate-reducing bacteria and aerobic and anaerobic methanotrophs (OMRMWR42 Holes). Sulphate reducing bacteria were also identified from lipid biomarkers and fatty acids in outcropping carbonate veins and travertine from the mantle section of the WTO43.

Isotopic evidence also supports current biotic activity at depth within the WTO. Measured fluid compositions, pH, and N and O isotope ratios, indicate that the amount of ammonium (NH4+) formed by NO3– reduction, and the proportion of NO3– that records a biotic isotopic signature, increase with depth40. Specifically, the authors observed a down-hole increase of the δ15N and δ18O values of NO3– (δ15N from 0‰–10‰ to > 20‰ and δ18O from ~ 20‰ to ~ 30‰) coupled with decreases in the concentration of NO3– from ~ 145 µM to ~ 1 µM40. These δ15N and δ18O trends were interpreted as record of nitrate reduction from biotic activity40. Bulk-rock carbon isotopes from the OmanDP Holes are consistently enriched in12C; the bulk-carbon δ13C values of material from holes BA3A and BA1B are − 9.54‰ ± 0.02‰ to − 5.78‰ ± 0.02‰, and − 15.9 ± 3.62‰ to − 7.24 ± 0.91‰, respectively. These low values are typical of biotic C fixation39.

However, low values of δ13C within the WTO are not unambiguously diagnostic of bioactivity because Fischer–Tropsch–type (FTT) abiotic synthesis of organic matter in hydrothermal systems can produce similar C isotope signatures to bioactivity44. In FTT reactions, organic molecules, such as fatty acids, are synthesised from abiotic H2 and CO2. Reaction kinetics are enhanced by catalysts, which may be metallic Fe, Ni, Co and Ru, Fe3+ or Zn oxides, and metallic sulphides45. For example, magnetite catalysed FTT reactions that formed organic molecules from abiotic H2 during serpentinisation experiments at 500 bar and 300 °C46. FTT reactions have been documented at a number of serpentinisation sites, including the Rainbow hydrothermal field 36°14’N in the Mid-Atlantic Ridge45 and the Kidd Creek mine, Ontario, Canada47.

On the more local scale, the morphology of nanoscale pores48 within chalcopyrite and copper from the WTO (Figs. 2 and 3B and C), is similar to that of pores within awaruite at the Chimaera seeps, Turkey49which are interpreted as a substrate for chemotrophic bacteria. The Chimaera seeps are a group of about 50 vents emitting mostly CH4 and are part of the Tekirova Ophiolite (TO)49. There, most of the CH4 is thought, based on C and H isotopic analyses of gases, to be produced abiotically by serpentinisation, with a minor biotic contribution50. While the WTO and TO are both continental ophiolites undergoing active serpentinisation that share textural features, there are also differences. The WTO does not host a hydrothermal system, WTO serpentinites are more serpentinised than TO serpentinites, and the WTO does not preserve the supra-subduction signature that has been observed within the TO and northern sectors of the Oman ophiolite51. Thus, despite similarities in the textural features and geodynamic setting of the Chimaera seeps and the WTO, similar features in the two locations may have formed by different processes, and the strong evidence of biotic activity may not relate directly to the observed C- and N-reduction and chalcopyrite oxidation.

Based on these observations, it is likely that redox reactions were mediated by biotic activity and abiotic processes in our samples, but their contribution is hard to distinguish unambiguously.

In our model, the reducing conditions associated with serpentinisation trigger desulphidation of chalcopyrite and reduction of Cu2+ to form secondary copper (Cu0, Fig. 4). Textural evidence excludes an alternative magmatic origin for the copper18. Magmatic copper typically forms inclusions in primary phases (e.g., olivine52), while secondary copper replaces magmatic sulphides, such as chalcopyrite, or forms isolated grains within serpentine veins18. Copper in our samples is porous, typical of secondary copper49and occurs within veins (aow3, Fig. 2e, f) or associated with ragged chalcopyrite and pentlandite partially replaced by magnetite (aow1 and 2, Fig. 2a–d). Moreover, thermodynamic models show that copper is stable relative to chalcopyrite at the conditions inferred for serpentinisation of the WTO at 250 °C18,35. It is possible that copper grains within the serpentine veins in sample aow3 record Cu addition during serpentinisation18 (Fig. 2). However, the average bulk rock Cu content of serpentinised WTO dunite and harzburgite is ~16 ppm51lower than the average upper mantle (25–30 ppm53), so Cu addition is considered minor or negligible and copper grains in serpentine veins most likely record limited length-scale Cu mobility during serpentinisation.

The presence of Ag on grain boundaries within Cu has not been reported previously, and Ag is a well-known catalyst for CO2 and O2 reduction in material science applications20. In other systems, CN-species catalyse incorporation of metal clusters in metalloenzymes54. Hence, it is interesting to speculate on the role of Ag in C- and N-cycling within the WTO (Fig. 3d). Partly serpentinised peridotite from the Nahlin Massif, Canada, hosts alloys of Au, Ag, Fe and copper55. There, it is thought that the precious and base metals originally hosted by magmatic sulphides transferred into native metals during serpentinisation, and that Au and Ag substitute for copper within the alloy crystal structure at concentrations of up to hundreds of ppm55. However, the techniques used for this previous research (EMPA and LA–ICP–MS) lack the spatial resolution necessary to confirm that elements are held within the crystal lattice. At the nanoscale resolution of APT, it is clear that Ag does not substitute for Cu in the lattice within the WTO sample. Instead, Ag decorates planes that form grain boundaries (Fig. 3d). At the inferred temperature of serpentinisation for the WTO (250 ℃35), Ag and Cu are immiscible at the Ag concentrations reported here (< 1 at%), so it is possible, not only that Ag facilitated redox reactions during serpentinisation, but that biotic activity contributed to the formation of Ag-bearing copper.

Subduction is thought to be the main mechanism of N transfer from surface systems to the mantle, although fluxes and the fate of N remain controversial17,56. The main vectors of N transfer are subducted marine metasediments and serpentinised oceanic lithosphere (57 × 109 mol yr–1 and 0.4 ± 0.2–14.7 ± 6.9 × 109 mol yr–1, respectively17,56). Literature suggests that most of the N in serpentinites is NH4+, adsorbed onto serpentine40,56. In contrast, our results reveal that N- and C-species are associated with alloys and sulphides, providing a previously unrecognised source of N in serpentinites that could be transported into the mantle within subducting slabs (Figs. 2 and 4).

These findings have implications for the preservation of N within the downgoing slab. Our samples, those of the Chimaera Seeps, Turkey, and other locations57 where microbial activity has been demonstrated, are from obducted ophiolites49. If fixation of N as CN-species only takes place after obduction, then subduction fluxes of these species may be limited because obducted serpentinites have a longer residence time within Earth’s crust than unobducted ocean floor serpentinites. However, the elevated N content of ocean floor serpentinites from Hess Deep, East Pacific Rise and Mid Atlantic Ridge (3.6–18 ppm) relative to depleted mantle (0.04–2 ppm) indicates that N is also added by serpentinisation on the ocean floor56. Microbial activity has been demonstrated in marine serpentinising environments16and whole-rock N contents of the WTO are similar to that of the ocean floor serpentinites40 so it is likely that biotically-mediated C- and N-reduction also occurs on the ocean floor, where it can contribute to subduction N fluxes.

A primary differentiating factor amongst theories for the origin-of-life relates to the speciation and availability of N7,9,13. It has long been suggested that serpentinising environments provided energy, nutrients, chemical gradients, and reducing conditions that supported the emergence of life3,41,49. Other studies conclude that kinetic constraints preclude an enzyme-free origin of life and that HCN formed by impactor vaporisation is a necessary and plausible progenitor to key organic molecules required for the first life9,13.

The role of catalysts and N are fundamental to resolution of these differences. For example, it has been proposed that the products of CO2 reduction in serpentinising environments are bound onto the surfaces of metals58providing a primitive N-free analogue of metabolic enzymatic processes11. However, direct observations have never been provided to support this proposal11. Indeed, most of the substantial body of knowledge that describes microbial life and diversity within serpentinites is without spatial context40,41. Our study provides that spatial and textural context, revealing a close spatial association between a metallic catalyst (copper), the precursor to that catalyst, in the form of chalcopyrite, a source of H2 in the form of unaltered olivine, and neoformed CN-species deposited on chalcopyrite.

Two factors impact consideration of our results in the context of life’s origins. First, the origin of life was, by necessity, abiotic, and the WTO serpentinising systems may not have been. Second, native Fe- and Ni-alloys, rather than copper, are the catalysts most commonly inferred to have contributed to the origin of life because of their presence as functional groups within primordial enzymes11. However, the presence of at least superficially similar compounds preserved within pyrrhotite-hosted fluid inclusions on the asteroid Ryugu14 and within aggregates mostly composed of hydrous silicates on Bennu15 suggest that these compoundscould form abiotically at low temperatures. Therefore, the components that supported emergence of life may have been linked by similar spatial relationships to those documented here.

We conclude that the previously unrecorded nanoscale CN species in sulphides and copper are related to serpentinisation, based on textural evidence linking copper formation to serpentinisation and the strong spatial association between copper and the CN compounds. While there is abundant evidence for biotic activity within the OmanDP Holes, formation of the CN species may have occurred biotically or biotically. Native Cu–Ag alloys catalysed CO2 and NO3– reduction, while pre-existing chalcopyrite may have provided a substrate for the microbiome. In contrast to previous work, our nanoscale observations show that Ag forms planes within Cu, rather than a lattice-substituted solid solution, with implications for catalytic facilitation of serpentinisation-related redox reactions. Furthermore, these previously undescribed C- and N-species provide a hitherto unconsidered vector for transport of N from the crust to the mantle by subduction, and a spatial context for origin-of-life theories.

Methods

Geological setting

The WTO represents a fragment of the Tethys Ocean floor that was subducted under the Arabian margin at about 95 Ma59. It consists of an amphibolite-facies metamorphic sole overlain by oceanic lithosphere (Fig. 159), . This work focuses on dunite and harzburgite from the cores drilled by the International Continental Drilling Program Oman Drilling Project (OmanDP34Fig. 1), Holes BA4A and BA3A, two of the seven drill sites selected to study links between weathering, serpentinisation systems, and the biosphere. Sample BA4A-79-2 was recovered from a depth of ~212 m, and sample BA3A-86-1 is from ~219 m (Supplementary Table 1).

Microscopy

Polished thin sections were inspected by optical and scanning electron microscopy (SEM). A Nikon Eclipse LV100PDL microscope was used in plane polarised light (PPL), cross-polarised light (XPL) and reflected light (RL). Backscattered secondary electron (BSE) images and qualitative energy dispersive X-ray (EDS) compositional spectra were acquired with an SEM Hitachi Tabletop Microscopy TM3030 and a Tescan MIRA 3 variable pressure-field electron microscope (VP-FESEM). Additional high-resolution BSE images were acquired using a Tescan Clara Field Emission Gun (FEG)–SEM. Instrument settings are provided in Supplementary Table 2.

ToF-SIMS

Samples for ToF–SIMS analysis (BA3A-86-1 and BA4A-79-2.1) were not C-coated to minimise C-contamination. ToF-SIMS maps were acquired using an Iontof M6. Imaged areas were pre-sputtered with the secondary beam using Cs+, to remove contamination on the surface. The instrument used Bi3+ as the primary ion source (liquid metal ion gun, LMIG). The mass resolution m/Δm for the identified ions and the experimental conditions are given in Supplementary Table 2. The software IONTOF Surface Lab v.7.3 was used to reconstruct the data. The data from aow1 was calibrated using Fe−, SiO2−, 63Cu−, 65Cu−, FeO−, SiO3−, CuS2; data collected from aow2 was calibrated with Fe−, SiO2−, 63Cu−, 65Cu−, FeO−, SiO3−, CuS−, CuS2−, while 37Cl−, 63Cu−, FeO3− and FeS2− were used as calibration for aow3. Maps were corrected for shifts in peak position during the analyses.

Atom probe tomography

The APT specimens (from BA4A-79-2) were manufactured using a Tescan Lyra 3 Focused Ion Beam Secondary electron microscopy (FIB-SEM), following the method described by27. Two APT specimens, M10 from aow4 and M19 from aow5, were analysed using a CAMECA LEAP 4000X HR (Supplementary Fig. 2, 3, Table 6). Experimental conditions are provided in Supplementary Table 2. Data processing and reconstruction were completed using the CAMECA software APSuite v.6.0. The mass–to–charge (m/q) spectra (Supplementary Fig. 3) were calibrated using the H+, 16O+, 58Ni2+, 65Cu2+, 58Ni+65Cu+, 195Pt2+, 107Ag+, 195Pt+ peaks.

The instruments are housed at the John de Laeter Centre or the School of Earth and Planetary Sciences, Curtin University, Western Australia.