Main

NASA’s OSIRIS-REx mission returned pristine samples of asteroid Bennu in September 2023 that exhibit extensive aqueous alteration1, consistent with spacecraft observations suggesting widespread hydrated minerals on the surface2,3. Alteration occurred on a parent asteroid, which was disrupted and portions reassembled to form rubble piles such as Bennu, for example, refs. 4,5. The dominant alteration phases (~80 vol%) are sheet silicates; other minerals that formed during aqueous alteration within Bennu’s parent body include sulfides, magnetite, carbonates, phosphates and late-stage evaporites1,6.

Similarly, the CI carbonaceous chondrites were mineralogically altered by fluid on their parent asteroid. However, these meteorites were also exposed to Earth’s moisture for decades, furthering mineralogic changes, for example, sulfate veins in CI chondrites of terrestrial origin7. Deconvolving parent-body mineralogic alteration from terrestrial weathering therefore adds to the challenge of understanding CI chondrite origins. Controlled return and curation of pristine samples from Bennu by OSIRIS-REx and asteroid Ryugu by JAXA’s Hayabusa2 mission8,9 offer an opportunity to study parent-body alteration free of terrestrial weathering. Here we describe the micro- to nanoscale mineralogy of Bennu samples (Supplementary Table 1), compare it to Ryugu samples and CI chondrites and discuss implications for Bennu’s protolith and origins of the most abundant alteration phases.

Survey of anhydrous minerals

Many submillimetre- to centimetre-scale Bennu particles have angular or hummocky shapes (Fig. 1a,b), respectively, similar to smooth and rough boulders observed on the asteroid (for example, Figs. 2 and 6 of refs. 1,10,11). Backscattered electron (BSE) imaging reveals carbonates (Fig. 1c), Fe-sulfides (Fig. 1d) and magnetite (Fig. 1e) in a submicrometre sheet-silicate matrix. X-ray diffraction (XRD) and phase mapping via electron microprobe analysis (EMPA) show that carbonates, sulfides and magnetite occur at modal abundances of 0.4–3.4%, ~3–8% and ~3–5%, respectively (Extended Data Fig. 1), comparable to values reported in ref. 1. Minor (1 to 10 vol%) to trace (<1 vol%) phases include anhydrous silicates, oxides, metal, phosphides, phosphates, carbon-bearing phases, nitride, sulfates, halides and fluoride (Supplementary Information).

Fig. 1: SEM and EMPA data from Bennu particles.
figure 1

a, BSE image of an angular particle from sample OREX-803095-0. b, BSE image of a hummocky particle from OREX-803096-0. c–e, BSE images of carbonate (OREX-803095-0), sulfide (OREX-803095-0) and magnetite (OREX-501030-0) grains. Yellow arrows point to pitting on grain surfaces. f, Fe+Co-Ni-S ternary diagram showing the composition of sulfides in samples OREX-803095-0, OREX-803096-0, OREX-803066-0, OREX-803067-0 and OREX-803070-0, together with reference sulfides. CI chondrite data from Bullock et al.12. *Stoichiometric sulfides include pyrrhotite Fe7S8 (4C), Fe9S10 (5C), Fe11S12 (6C) and Fe10S11 (11C); troilite (FeS) and pyrite (FeS2) for reference. g, BSE image of olivine from OREX-501019-0. h, Secondary-electron image of pyroxene from OREX-501059-0. i, BSE image of spinel from OREX-501019-101.

Full size image

Textures, sizes and compositions of these components are varied. Carbonates range from submicrometre-sized to ~100 μm, with mostly subhedral morphologies. Carbonates can exhibit pitted and embayed surfaces (Fig. 1c). Measured compositions span from nearly pure calcite to dolomite and Fe–Mn-rich magnesite, with rare sodium carbonate6. Fe-sulfides range from nanometres to hundreds of micrometres and exhibit euhedral to anhedral morphologies. Many sulfide grains have pitted surfaces (for example, Fig. 1d). Sulfides are predominantly pyrrhotite with trace pentlandite and cubanite. Pyrrhotite compositions measured by EMPA and scanning electron microscopy–energy-dispersive X-ray spectrometry (SEM-EDS) are close to Fe7S8 (the 4C polytype). Nickel contents vary from <1 to ~11 at%, extending towards pentlandite in (Fe+Co)-Ni-S ternary space (Fig. 1f). The Fe/S ratios of pyrrhotites for both angular and hummocky particles are on average 0.85, comparable to averages measured in CI chondrites12,13 and Ryugu14. Magnetite occurs in multiple morphologies1, dominantly plaquettes and framboids in vugs, pores and veins11, with sizes from micrometres to tens-of-micrometre assemblages (for example, Fig. 1e). Spherulitic magnetite also occurs. Like carbonates and sulfides, some magnetite grains have pitted surfaces (for example, Fig. 1e). The composition of the magnetite is near endmember Fe3O4 with minor solute elements Si, Mg, Ca, Mn, Cr and V (Supplementary Table 2).

Pre-accretionary dust

Remnants of material formed before parent-body accretion occur at minor to trace levels in Bennu samples. The most abundant of these is olivine (for example, Fig. 1g), at up to 3 vol%. Olivine occurs as clusters or isolated anhedral grains ranging from submicrometre to ~40 μm in size and exhibits angular or partially rounded shapes. Most olivines are forsteritic (Fo98–100) with a minor fraction being more fayalitic (Fo<90). The Mn/Fe ratios of Fe-rich olivines range from ~0.01 to 0.03, comparable to chondrule olivine in the most primitive chondritic meteorites15. Many olivine grains have textures, minor element abundances and cathodoluminescence (CL) features suggestive of amoeboid olivine aggregates (AOAs16; Extended Data Fig. 2), an observation supported by oxygen isotope data17. Some olivine grains are toroidal and rarely enclose Fe-Ni metal. Larger (hundreds of micrometres) assemblages composed of forsterite with radial textures sometimes contain partially rounded surfaces (Extended Data Fig. 3) and are similar to cryptocrystalline forsterite chondrules18,19. Despite this example, chondrules and chondrule pseudomorphs are exceptionally rare (trace components) in Bennu samples studied to date. Pyroxenes, both enstatite and diopside, are a minor to trace component of the Bennu samples and occur as ≲40 μm isolated grains, often corroded (for example, Fig. 1h). Refractory minerals typical of those in calcium–aluminium-rich inclusions (CAIs) occur in the sample, but they too are exceptionally rare. Mg-Al spinel grains ≲10 μm across occur as isolated clusters in the matrix with irregular shapes and corroded textures (Fig. 1i). Rare spinel-hibonite assemblages are observed (Extended Data Fig. 2). Minor hibonite, ~10 μm in length, occurs with a Mg-Al spinel grain. Schreibersite occurs with anhydrous silicates. Other refractory phases include presolar grains that are present at 29, 11 and 4 ppm levels for SiC, graphite and O-rich grains, respectively, similar to those in CI chondrites17.

Matrix mineralogy

Particles are mostly (≳80 vol%) composed of sheet silicates, both fine grained (tens of nanometres) and coarse grained (hundreds of nanometres; Fig. 2a,b). Transmission electron microscope (TEM) analyses of focused-ion-beam (FIB) cross sections and fine powders show that coarse-grained sheet silicates occur in radial assemblages containing a nucleus with fibres extending outward several hundred nanometres (Fig. 2c). Many fibres are sinuous and terminate into circular shapes in contact with fine-grained sheet silicates (Fig. 2c). Other coarse-grained sheet silicates occur in subhedral bundles containing (sub)parallel layers hundreds of nanometres across (Fig. 2d,e). High-resolution TEM (HRTEM) shows that many coarse-grained sheet-silicate bundles contain local long-range order (Fig. 3a,b) with domains of serpentine and saponite intergrowths (0.7- and 1.0-nm d spacings, respectively) tens of nanometres thick (Fig. 3c). Some areas of serpentine fibres are curled with chrysotile-like structures (Fig. 3d, arrowhead). Selected-area electron-diffraction (SAED) patterns show that fine-grained and some coarse-grained sheet silicates are poorly crystalline to amorphous (Extended Data Figs. 4–6).

Fig. 2: TEM data on FIB sections from matrix.
figure 2

a,b, STEM-HAADF (high-angle annular dark field) images of FIB sections extracted from OREX-803095-100 and OREX-803096-100, respectively. c, Bright-field (BF-TEM) image of the dashed box area in panel b, showing the coarse- and fine-grained sheet silicates (C.G. and F.G.). d,e, BF-TEM images of OREX-501005-100 showing fine-grained sheet silicates and their relationship to the coarse-grained sheet silicates. Sulfides (opaque grains), including pyrrhotite (Po in c) and pentlandite, occur in the fine-grained sheet silicates.

Full size image
Fig. 3: TEM data on FIB sections from matrix.
figure 3

a, BF-TEM and HRTEM image of OREX-501005-100 showing a cluster of coarse-grained sheet silicates. b, Close-up of the red box area in a. c,d, HRTEM images of OREX-501017-101 show local order of the serpentine (serp) and saponite (sap) intergrowths and curling of the serpentine structure into a chrysotile-like phase (arrowhead in d). The (001) planes are marked in c along with their characteristic d spacings.

Full size image

Sheet silicates are Mg rich with minor Al and Fe intermediate between saponite and serpentine solid solutions in Mg-Fe-(Si+Al) ternary space (Fig. 4a and Extended Data Fig. 4f–k). Several coarser-grained sheet silicates from a hummocky particle have higher Fe, extending towards greenalite or cronstedtite compositions, the latter of which is an abundant alteration product in carbonaceous Mighei (CM) chondrites20. Compositions of sheet silicates in Bennu grains overlap similar phases in the Orgueil CI chondrite and Ryugu (Fig. 4a). X-ray absorption near-edge structure (XANES) analysis shows that Fe in the sheet silicates occurs as both Fe2+ and Fe3+, with Fe3+/ΣFe varied between 0.5 and 0.6 among grains (Fig. 4b,c).

Fig. 4: Composition of sheet silicates.
figure 4

a, Mg-Fe-Si+Al ternary diagram. Saponite and serpentine solid-solution lines are displayed together with endmember serpentine compositions (arrows). The compositional fields for sheet silicates measured in Ryugu and Orgueil are shown for reference; data from refs. 8 and 42, respectively. b, Fe L2,3 XANES map of FIB section OREX-803031-101 superimposed on a STEM-HAADF image. c, XANES Fe L2,3 edge spectra of optical density (OD) vs. electron volts (eV) showing that the matrix contains a mixture of Fe2+ and Fe3+. Numbered spectra correspond to numbered locations in b. Areas in yellow contain more Fe3+ than Fe2+; those in purple contain more Fe2+ than Fe3+. Faint yellow lines in b suggests a possible boundary between reduced and oxidized matrix.

Full size image

Fine-grained sheet silicates are decorated with Fe-sulfides hundreds of nanometres to several micrometres across (for example, Fig. 2c–e). SAED patterns show that most sulfides are 4C pyrrhotite polytypes, but 5C also occurs (Fig. 5a,b,e and Extended Data Fig. 7). Similarly, non-integral types were identified in CM chondrites21. The majority of nanoscale pyrrhotite grains are subhedral to anhedral with a very minor amount being euhedral (for example, Fig. 5a–c, inset for Fig. 5f,g). Minor pentlandite occurs in fine-grained domains (Fig. 5a), and we observe intergrowths of pyrrhotite and pentlandite (Fig. 5c–e). Sheet silicates occur both parallel and normal to sulfide surfaces (for example, Fig. 5f,g). We find inclusions of sheet silicates within pyrrhotite (Extended Data Fig. 8). Minor olivine and pyroxene are surrounded by fine- and coarse-grained sheet silicates (Extended Data Fig. 7).

Fig. 5: TEM data on sulfides.
figure 5

a, Pentlandite (Pn) and pyrrhotite (Po) occur adjacent to one another in the matrix of OREX-803096-100. b, Pyrrhotite occurs in the matrix adjacent to a nanoglobule (Ng) in OREX-803095-101. c,d, BF-TEM image of OREX-501005-100 showing pyrrhotite in the matrix (c) with a corresponding EDS map in red, green and blue (Fe, S and Ni, respectively) showing pentlandite exsolution (purple areas) (d). e, HRTEM image from the area in the white box in (c) showing the structural relationship between the pyrrhotite and pentlandite. The (111) planes of pentlandite are parallel to the (001) planes of pyrrhotite, consistent with exsolution during slow cooling from higher temperatures69. f,g, STEM-HAADF image (inset; scale bar is 500 nm) of a euhedral pyrrhotite grain from OREX-501044-100 near [001] projection, with adjacent sheet silicates normal (f) and parallel (g) to its surface.

Full size image

Carbon-bearing material in the matrix includes carbonates and insoluble organic matter, with carbonaceous nanoglobules being the most obvious and discrete morphological type of the latter. The carbonates occur as abundant dolomite, including veins that cross cut the sheet-silicate matrix, and Fe–Mn-rich magnesite reaching several hundred micrometres (Fig. 6a). Some grains display a core-shell texture of dolomite-magnesite or the inverse. Panchromatic CL imaging of carbonate grains reveals distinct epitaxial overgrowths (Fig. 6b,c) from multiple precipitation and crystallization events. Calcite occurs in open pores and vugs within the matrix (Fig. 6d) as small (<10 µm) euhedral grains associated with magnetite framboids and rare Fe–Mn-rich magnesites. CL analyses (Fig. 6e) show that some calcite grains are unzoned and Mn-poor. At the TEM scale, we also identify core-shell carbonate with a siderite core surrounded by dolomite and Fe–Mn-bearing magnesite (Fig. 6f–h). Carbonaceous nanoglobules range from tens of nanometres to >1 μm across (Fig. 6i–k and Extended Data Fig. 9). They contain O and minor N and S. TEM imaging shows that the nanoglobules occur within the fine-grained sheet-silicate matrix (Figs. 2, 5 and 6i,j and Extended Data Fig. 9), but some nanoglobules encapsulate sheet silicates (Fig. 6i,j). XANES reveals nanoglobules and diffuse carbon interwoven with sheet silicates in varied functional groups (Fig. 6k,l). Scanning TEM (STEM)–EDS and XANES measurements show that C is on average more abundant in the fine-grained than the coarse-grained sheet silicates (Fig. 6j,k).

Fig. 6: SEM, TEM and XANES data on C-bearing materials.
figure 6

a, High-resolution Ca and Mg X-ray map of OREX-800045-103 with larger dolomite (D) and Fe–Mn-rich magnesite (M) crystals in fine-grained sheet-silicate matrix. Some grains display a core-rim texture, with either dolomite or Fe–Mn-rich magnesite forming the core and the other mineral overgrowing (red arrows). b, BSE image of a large carbonate from OREX-800045-103. c, Panchromatic CL image of the same grain revealing a core-shell texture, with Fe–Mn-rich magnesite (M) non-luminescent crystals forming the core and epitaxial overgrowths of dolomites. d, High-resolution X-ray chemical map of a micrometre-sized open pore in the matrix of OREX-800045-106 with several euhedral calcite grains, magnetite framboids of various size and rare Fe–Mn-rich magnesite. A thin serpentine layer lines the pore. e, Panchromatic CL image of the same area showing the net luminescence of the calcite grains. f–h, BF-STEM and EDS maps of OREX-501005-101 revealing a zoned euhedral magnesite aggregate containing a siderite core and chemical zoning. i,j, BF-TEM image and EDS map of a C nanoglobule in OREX-501005-101 surrounded by and including sheet silicates. k, C-K XANES map of the C nanoglobule and surrounding sheet silicates in i,j. l, XANES spectra of C-K edge showing that C is spatially correlated with fine-grained sheet silicates and nanoglobules. Numbered spectra correspond to areas indicated in k. Varied functional groups are identified in the C-K edge. The peak at 284.8 eV is characteristic of aromatic C=C bonds. The peaks at 286.5, 287.4 and 288.3 eV are due to various functional groups, probably bonded to O and H, for example, enol, phenol, vinyl ketone, alkyl and carboxyl.

Full size image

Conditions of parent-body alteration

Conditions of aqueous alteration can be estimated from sulfide and carbonate compositions. Isothermal sections of the Fe-Ni-S ternary system22,23, coupled with compositions of co-existing sulfides in CI and CM chondrites, constrain temperatures of crystallization and last equilibration12,24,25. Pyrrhotite compositions plot closest to the pyrrhotite (4C)–pentlandite tie line in the 25 °C isothermal section (Extended Data Fig. 10) and at higher S concentrations relative to the tie line in the 100–135 °C isothermal section. These data suggest alteration at ~25 °C, with an upper limit of ~100 °C, consistent with the co-occurrence with sheet silicates and carbonates. Similar temperatures were derived for Ryugu, with sulfide compositions indicating 25 °C and oxygen isotopic compositions of dolomite and Fe-rich magnesite indicating temperatures up to 100 °C (refs. 26,27,28). In comparison, temperatures of 50 to 150 °C are estimated for CI chondrites based on thermodynamic modelling29.

Studies of CI chondrites have argued for closed-system alteration30,31,32 based on a lack of depletion or enrichment of fluid-mobile elements. Modelling by ref. 31 assumed 150 °C (ref. 29) and calculated water-rock ratios of 0.5 to 1 in a closed system. In comparison, carbonate veins in boulders on Bennu33 and enrichment of fluid-mobile elements in Bennu samples1 suggest open-system alteration. Laboratory alteration experiments simulating hydrothermal alteration at terrestrial mid-ocean ridges produce serpentine at water-to-rock ratios of ~50 to 12,000, temperatures of 170 to 300 °C, and pressures of 25 to 50 MPa (refs. 34,35). In these experimental simulations of olivine dissolution, temperature may be the most important of these variables because closed-system experiments at 90 °C for 211 days produced very limited reactions in mafic silicates, perhaps a result of a high nucleation barrier36. In terrestrial systems, pH, water activity, temperature and interfacial surface area are the major factors controlling olivine dissolution37.

Mineral relationships in the Bennu samples could be explained by an open or closed thermodynamic system. Sulfides and sheet silicates formed simultaneously, as evidenced by inclusions of sheet silicates in sulfides and sheet silicates conformal and normal to sulfide surfaces (Fig. 5 and Extended Data Fig. 8). Open-system alteration might allow changing S concentration to produce sulfide-bearing and sulfide-poor phyllosilicates (Fig. 2), although requiring submicrometre-scale fluid-flow regimes. Alternatively, differences in sulfide abundance within phyllosilicates might result from heterogeneous precursors altered by a common fluid in either an open or closed system. Both types of system allow an evolving fluid in alkalinity and pH, evidence for which is provided by Bennu samples6 and thermodynamic models of CI chondrites29,32, which suggest that the fluid evolved from neutral to alkaline with pH of 9 to 10. Evidence for fluid evolution is found in etch pits and embayments on sulfides and carbonates (Fig. 1c–e), suggesting dissolution under conditions that differed from those present during their precipitation. Similar textures were observed on pyrrhotites in the Orgueil and Ivuna CI chondrites and Ryugu samples12,38. Sheet-silicate reprecipitation events probably also occurred, particularly in void spaces (Extended Data Fig. 8), explaining the diversity of sheet-silicate textures.

Comparison to Ryugu and CI carbonaceous chondrites

Mineralogically, Bennu and Ryugu are similar. Ryugu samples are dominated by fine-grained serpentine and saponite8,27,39. Samples from both asteroids contain comparable types, amounts and compositions of sulfides, magnetite and carbonate. Similar assemblages of minor and trace minerals and phases, including late-stage, sodium-rich phases formed by evaporation6,40, are observed in both asteroids, with differences probably reflecting the limited mass of samples studied. CAIs and chondrules, both unaltered and altered pseudomorphs, are rare in Ryugu samples8,41.

Compared to chondritic meteorites, Bennu and Ryugu are most like the CI types1,8,27,39. CI chondrites are also dominated by Mg-rich serpentine and saponite decorated with carbonates, sulfides and magnetite42. The composition and structure of sheet silicates in Bennu are very similar to those that occur in the Orgueil CI chondrite (Fig. 4a). Chondrules and CAIs are rare in CI chondrites43. The prevalence of Ca-sulfates and ferrihydrite in the matrix of CI chondrites42,44,45 distinguishes them from Bennu and Ryugu particles. We have not observed ferrihydrite in the Bennu sample and it was not reported in Ryugu, suggesting that terrestrial weathering of sulfides formed the sulfates44,45 and ferrihydrite46.

CI chondrites are rare meteorites, with only six unheated examples (falls Orgueil, Ivuna, Alais, Tonk, Revelstoke, and a recent find Oued Chebeika 002) with a combined mass of ~21 kg (ref. 47). That the first two sample-return missions to near-Earth carbonaceous asteroids returned material similar to CI chondrites implies that the terrestrial meteorite collections are biased, that is, there may be many more chemically primitive, volatile-rich asteroids than suggested by what has been recovered on Earth. The underrepresentation of these materials on Earth is related to the break-up of particular asteroid families48 but also probably due to their friable nature precluding them from surviving atmospheric entry, as was suggested for Bennu’s lowest-thermal-inertia materials49. Those that do survive are subjected to terrestrial weathering that further degrades them, highlighting the importance of controlled return and curation of pristine samples from asteroidal bodies.

Bennu’s protolith

In the conventional paradigm, parent bodies of carbonaceous chondrites (and by extension, Bennu) accreted in the protoplanetary disc from a mixture of chondrules, refractory inclusions, anhydrous silicates, oxides, sulfides, metal, carbonaceous matter and ices, for example, ref. 50. Heating from accretion, radioactive nuclides or impact, for example, refs. 51,52, melted the accreted ice, resulting in a hydrothermal system that drove secondary alteration. In the most aqueously altered chondritic materials (petrologic types 1 and 253), aqueous fluid reacted with initially accreted components, forming sheet silicates, sulfides, magnetite, carbonate, and other minor phases observed in these meteorites and samples from Bennu and Ryugu.

Most chondrites form from a chondrule-rich and occasionally refractory inclusion-rich material50. However, two lines of evidence argue against such a model for Bennu. First is the rarity of observed chondrule pseudomorphs in the Bennu samples analysed to date (however, consider Extended Data Fig. 3). Several chondrule-like objects and a single, 30-µm chondrule pseudomorph were observed in Ryugu samples8. In contrast to Bennu, Ryugu and CI chondrites, the CM1 chondrites, which formed at similar water-to-rock ratios (0.2 to 0.7) and temperatures (20 to 200 °C), contain abundant chondrule pseudomorphs54. Thus, if chondrules and CAIs accreted to the Bennu precursor, they were probably a minor component. The second line of evidence is bulk chemical data showing that Bennu has an unfractionated composition1, whereas CM chondrites show enrichments in refractory elements and depletions of volatile elements that are consistent with the 20% of chondrules and CAIs with which they accreted55,56.

An alternative model involves accretion of chondrule- and CAI-poor nebular dust composed of nanometre- to micrometre-scale silicates. The major difference in this model is the relative abundance of fine- and coarse-grained materials. Given the degree of aqueous alteration experienced by Bennu samples, the exact crystalline nature of this nanometre- to micrometre-scale fraction is poorly constrained. One possibility is alteration of an initial mixture of nanometre- to micrometre-scale crystalline and amorphous silicates. Although we have not yet observed amorphous material in Bennu, clasts dominated by amorphous silicates occur in Ryugu57 and are common in mineralogically primitive meteorites58,59,60.

A mixture of pre-accretionary crystalline and amorphous silicates might explain features observed in Bennu, particularly the coexistence of coarse-grained sheet silicates with few or no sulfide inclusions and fine-grained sheet silicates rich in sulfide inclusions at the micrometre scale (Fig. 2). Many amorphous silicate grains such as those in interplanetary dust particles contain nanometre- to micrometre-scale sulfide inclusions61,62 such as those observed in fine-grained sheet silicates of Bennu. Amorphous silicates rapidly hydrate at room temperature under alkaline conditions63, such as those inferred for Bennu’s parent body6. In such a model, the coarse-grained sheet silicates, which lack sulfide inclusions, might form from crystalline silicates, whereas the fine-grained sheet silicates, which contain sulfide inclusions, could have formed from amorphous silicate grains with embedded pyrrhotite, similar to those commonly found in anhydrous interplanetary dust particles61,62,64 that originated in the outer reaches of the solar system65. Such grains would have to be consumed in the alteration and reprecipitated as secondary sheet silicates and sulfides.

Taken together, the mineralogical evidence presented here suggests that Bennu’s protolith sampled an inventory of starting material from the protoplanetary disc that was poor in chondrules and CAIs, rich in nanometre- to micrometre-scale silicates and contained AOAs. Thus, the parent body of Bennu, along with those of Ryugu and CI chondrites, was richer in matrix components and poorer in refractory inclusions than those of type 2 or 3 carbonaceous chondrites—or even CM1 and carbonaceous Renazzo (CR1) chondrites (≲20 vol% and 50–60 vol% chondrules and refractory inclusions, respectively56), which experienced similar degrees of aqueous alteration to Bennu and CIs but contain chondrules. A growing body of data suggest that the materials that comprise Bennu, Ryugu8 and CI chondrites55 may be transitional in the nature of the initially accreted material between CM chondrites and cometary objects such as Wild 2, which contains chondrule glass and rare CAIs66,67, possibly pointing to an origin in the outer Solar System, consistent with both isotopic and organic evidence17,68.

Methods

Several different spatially resolved analytical techniques were used for mineralogic analysis of the samples. Briefly, initial inspection of aggregate (bulk unsorted) material was performed using visible light microscopy. Some particles were sufficiently large that they could be separated based on their morphological type: hummocky or angular (1,11). In such cases, polished sections were prepared. However, small (<1 mm) particles precluded definitive identification of morphologic type and so no separation was made. The particles were prepared in multiple ways including polished sections, ultramicrotome mounts or crushing and drop-casting onto TEM grids. We used XRD, SEM, CL spectroscopy and EMPA to determine, at the millimetre to micrometre scale, the mineralogy, textures, spatial relationships and modal mineral abundances of the particles. Once that contextual information was acquired, we used focused-ion-beam scanning electron microscopy (FIB-SEM) to extract and prepare, in situ, electron-transparent cross sections from fine-grained areas of tens of particles for analysis using TEM. Detailed information on crystal structure, including long- and short-range atomic order, crystal chemistry and spatial relationships, was provided by TEM imaging (including HRTEM and STEM), SAED, EDS and electron energy-loss spectroscopy (EELS). XANES was used to assess transition-metal oxidation states and functional chemistry.

For all laboratories, samples were stored in desiccator cabinets when not being measured. To avoid terrestrial alteration via exposure to ambient moisture, most desiccator cabinets incorporate a nitrogen purge to maintain ultra-dry conditions.

Visible light microscopy

University of Arizona

Particles received from the aggregate material were imaged and their morphologies assessed using optical microscopy. Reflected light images were acquired using a Keyence VHX-7000 digital microscope. Particle shapes that exhibited angular and hummocky morphologies were extracted from the allocated material and prepared as 1-inch round polished sections. Six particles of each type were embedded into 1-inch round and 1/8-inch tall epoxy plugs using with Buehler EpoxiCure 2 resin. The surface of the plugs was impregnated with superglue and polished dry or with ethanol if necessary.

Scanning electron microscopy

NASA Johnson Space Center

Samples were characterized using a JEOL 7600F FE-SEM equipped with a 170-mm2 silicon drift detector (SDD) type Oxford Instruments Ultim Max EDS detector. The particle was attached to an Al cylinder SEM mount using double-sided C tape. After initial documentation by optical microscopy, the sample was sputter coated with ~5 nm of C to assist with charge dissipation during SEM analysis. The Oxford AZtec ‘Point & ID’ programme was used for the acquisition of images and point spectra. Characterization of regions of interest was performed at an accelerating voltage of 15 kV using both secondary electron (SE) and low-angle backscattered electron imaging modes. EDS spectra were acquired at 15 kV with acquisition times ranging from 20 to 200 s with an incident beam current of ~ 900 pA.

University of Arizona

Polished sections were coated with a thin layer (0.1 nm) of carbon to mitigate charge build-up in the electron microscope. SE and BSE images were acquired using a Hitachi S-4800 SEM located in the Kuiper-Arizona Laboratory for Astromaterials Analysis (K-ALFAA). The Hitachi S-4800 cold-field emission gun (cold FEG) is a 0.5 to 30 keV SEM equipped with SE and BSE imaging detectors and an Oxford Instruments Aztec Live/x-stream/Ultimax 170 SDD energy-dispersive X-ray spectrometer (EDS). The compositional heterogeneity of the particles was assessed through EDS mapping.

Electron microprobe analysis

University of Arizona

BSE images, element maps and quantitative compositional analyses of the particles were acquired using the Cameca SX-100 Ultra electron microprobe in the K-ALFAA. X-ray maps and BSE images were run at 15 kV and 20 nA. Quantitative analyses of silicates, sulfides and oxides were run using a focused beam at 15 kV, 20 nA, 20 s peak time and 10 s on each background. A 2-μm defocused beam size, lower beam currents and shorter count times were used for phosphate and carbonate analyses to minimize possible beam damage effects. Phosphate analyses were performed at 15 kV, 8 nA, 20 s peak and 10 s background; carbonates at 15 kV, 4 nA, 10 s peak and 5 s background. Well-characterized natural and synthetic materials were used as standards.

Phase maps of polished sections were created from the element maps using XMapTools70,71. The pixels representing individual phases, that is, carbonates, sulfides, magnetite and ‘matrix’, were defined with varied polygons. The algorithm in XMapTools (classification tree or k-means) was trained to recognize those pixels and produce a phase map based on counts. Matrix was defined as any area in the maps with particle sizes <1 μm. Grain sizes below ~1 μm are generally not detected by EMPA, so nanoscale grains, for example, sulfides, that occur within it will occur in the maps. In comparison, such nanoscale grains will diffract and appear in the XRD pattern. The differences in the modal abundances, as determined by EMPA and XRD, can be attributed to the differences in detectability of grain sizes.

Cathodoluminescence spectroscopy

Université Cote d’Azur and Observatoire de la Côte d’Azur

The high-resolution cathodoluminescence facility is mounted on a FEG‐SEM JSM 7000F JEOL. The MonoCL4 GATAN monochromator is equipped with both a high-sensitivity array detector and a high-sensitivity photomultiplier, suitable both for hyperspectral analyses and panchromatic and monochromatic imaging (CRHEA Valbonne, France). Collection of the integrated luminescence extracted from the sample is achieved using a paraboïdal mirror placed above the sample. The electron beam is directed perpendicularly to the sample surface through a hole in centre of the mirror. For panchromatic imaging, recording is achieved using the high-sensitivity photomultiplier detector. The electron-beam current used in the study ranges typically from 1 to 4 nA, and the voltage beam has been fixed at 5 keV. For this accelerating voltage, the emitting volume expands up to 230 nm below the bombarded sample surface and around to 200 nm sideways, assuming a 100-nm graphite coating thickness on Bennu particle thin sections. To avoid as much as possible the CL ‘hotspot’ effect, particularly effective at low magnification, the recording was made with the sample placed lower than the focal plane of the paraboïdal mirror for magnifications less than ×500.

Focused-ion-beam scanning electron microscopy

University of Arizona

Electron-transparent cross sections were created for TEM analysis using the ThermoScientific Helios G3 FIB-SEM located in the K-ALFAA. All sections were extracted from varied regions of matrix within the particles. BSE and SE images were acquired before and after sectioning. The FIB sections were thinned to electron transparency using previously described methods (for example, refs. 72,73,74,75). Briefly, a 12-μm wide × 4-μm tall protective capping layer of C was deposited on top of areas of matrix. The lamellae were created using a standard stair step, extracted in situ and thinned to electron transparency at 30 keV and currents ranging from 2.5 to 0.8 nA.

University of California at Berkeley

Bennu particles were placed on PELCO carbon conductive tabs on an aluminium SEM round. FIB sections were extracted using an FEI Helios G4 UX dual-beam FIB-SEM at the Molecular Foundry at the Lawrence Berkeley National Laboratory. Coarse milling was done using a Ga+ ion beam at voltages between 16 and 30 keV, and polishing was done at various voltages down to 1 keV. FIB sections were made in various thicknesses from <100 to 600 nm, depending on the targeted experiment. Thicker sections were used for Fe-L X-ray spectroscopy and tomography, whereas thinner sections were used for C-K X-ray spectroscopy and TEM.

NASA Johnson Space Center

FIB sections were prepared from particles dispersed on conductive carbon dots on Al SEM pin mounts using a FEI Quanta3D600 dual-beam FIB. For most sections, we used electron-beam deposition to deposit ~0.5–1 μm of carbon, followed by 2–3 μm of ion beam-deposited carbon for the capping layer. Milling was done in steps using a Ga+ ion beam initially at 30 kV, followed by milling at 16 kV and final thinning at 5 kV until the sections were ~100 nm thick. We used ion beam-deposited Pt to weld the section to the Cu half grids.

Transmission electron microscopy

University of Arizona

Characterization of the FIB sections was performed using the 200 keV Hitachi HF5000 STEM, located in K-ALFAA. The HF5000 is equipped with a cold FEG; a third-order spherical-aberration corrector for STEM mode; bright-field (BF), dark-field (DF) and SE STEM detectors; an Oxford Instruments X-Max N 100 TLE EDS system with dual 100 mm2 windowless SDDs providing a solid angle (Ω) of 2.0 sr; and a post-column Gatan Quantum EELS. TEM images and electron-diffraction patterns were acquired with a Gatan OneView 4,096 × 4,096-pixel CMOS camera.

The sample was imaged in and compositionally mapped in STEM mode (converged beam) using EDS. The EDS spectrum images were acquired with a 136 pm probe size at 512 × 512 pixels over a 20 keV energy range with 2,048 channels, a process time of three (roughly equivalent to a time constant) and a frame time of 8 μs. For quantification of sheet-silicate composition, we acquired spectrum images over large fields of view containing multiple phases. However, we selectively quantified the smaller domains containing only fine- or coarse-grained phyllosilicates thereby eliminating any contribution of nanoscale sulfides or magnetites to the quantified composition. Selected regions of interest were extracted from the spectrum images as summed spectra containing only sheet-silicate compositions and those were quantified.

Areas of the FIB section were analysed in TEM mode (parallel illumination) for grain size and crystallinity. BF-TEM images and SAED patterns were acquired from select regions of interest. The SAED patterns were measured using Crystallographic Image Processing Software Package76 and by hand in Adobe Photoshop based on calibrated camera constants.

University of California at Berkeley

TEM analysis was done on an FEI TitanX microscope at beam energies between 80 and 300 keV. STEM/EDS maps were typically acquired at beam currents <0.25 nA using a 0.6-sr Bruker quad SDD. To control for volatilization of light elements, especially hydrogen and loosely bound oxygen, maps were acquired sequentially for periods of several minutes to >1 h and then combined using Python.

Goethe University

TEM samples were prepared by crushing the grain, dropping ethanol onto the resulting powder and touching a TEM copper mesh grid with a lacey carbon support to the suspension until all material was picked up. TEM data were collected in the Schwiete Cosmochemistry Laboratory at Goethe University on a ThermoScientific (TS) Talos F200-X G2 S/TEM, operated at 200 kV. TEM images and SAED patterns were acquired with a TS Ceta-S 4k × 4k 16 M camera. Image magnification and camera constant calibrations were performed with a Cross Grating 3 mm S106 sample from AGAR S106. STEM-HAADF images were collected with a convergence semi-angle of 10.5 mrad and an inner collection semi-angle of 58 mrad. EDS data were collected using four windowless TS EDS silicon drift detectors. TS Velox was used for all data processing.

NASA Johnson Space Center

FIB sections were analysed using a JEOL 2500SE 200 kV STEM equipped with BF, DF and SE detectors for imaging in STEM mode and a Gatan OneView 4,096 × 4,096-pixel CMOS camera for HRTEM imaging and electron diffraction. The JEOL 2500SE is equipped with a JEOL 70 mm2 SDD for EDS analyses and spectrum imaging and a Gatan Tridiem GIF for EELS analyses. EDS data were quantified using k-factors from well-characterized standards and a Thermo System7 analytical system. Spectrum images (typically 256 × 200 pixels) were obtained by rastering a 2-nm incident probe over the region of interest with a 50-μs dwell time per pixel. Successive frames were accumulated until counting statistics were better than 1% for major elements.

Synchrotron spectroscopy

University of California at Berkeley

Synchrotron studies were carried out at Scanning Transmission X-ray Microscopy beamline 5.3.2.2 at the Advanced Light Source at Lawrence Berkeley National Laboratory. Stacks were acquired from FIB sections using an X-ray beam focused using a Fresnel zone plate down to several tens of nanometres and rastering the sample at various beam energies to produce data cubes with two spatial dimensions and one energy dimension. The edges studied included C-K, O-K and Fe-L. Spatial steps were typically on the order of the spot resolution, and energy steps ranged from 0.1 to several eV depending on the expected presence of features in the spectrum.

X-ray diffraction

NASA Johnson Space Center

Sample OREX-500005-0 (~88 mg) was characterized using XRD at NASA Johnson Space Center. The sample was initially imaged in a glass vial using an optical microscope and found to consist mainly of fine (<100 μm) particles, although several intermediate particles up to ~500 μm in size were also observed. It was then poured onto a zero-background substrate and gently pressed to ensure the largest sample surface area was exposed to the X-ray beam. Analyses were conducted using a Malvern Panalytical X’Pert Pro scanning XRD instrument. XRD patterns of the sample were collected under ambient conditions using Co Kα radiation from 2–90° (2θ) with a step size of 0.006°. An initial measurement (~90 min) was used to identify the main phases, followed by a longer measurement (~17 h) to achieve a good signal-to-noise ratio. The 2θ accuracy and the baseline were verified using several standards including silicon (NIST 640f), LaB6 (NIST 660c), novaculite and a blank substrate. Final mineral identification and quantitative phase analysis (Rietveld refinement) were performed using the X’Pert HighScore software package.