Gradual Destruction of Refractory Stardust in Alkali Brines on Asteroid Bennu

Abstract

Carbonaceous asteroids preserve presolar grains—refractory stardust that predates the Solar System—recording both their stellar origins and subsequent alteration on their parent asteroids. Here we examine the distribution of presolar grains and isotopically anomalous organic matter in mottled particles from samples returned by the OSIRIS-REx mission from asteroid Bennu. Although the mottled samples contain organic matter with carbon and nitrogen isotopic compositions similar to those of other Bennu samples, they lack presolar silicate or oxide grains and show a markedly reduced abundance of presolar silicon carbide. In this work, we show that these findings are consistent with the slow destruction of silicon carbide grains during late-stage alteration by a low-temperature, oxidizing, alkaline brine. These brines, coinciding with evaporite salt formation, selectively erased chemically robust presolar phases while preserving isotopically anomalous organic matter. This shows that evolved aqueous fluids can fundamentally alter the presolar record of primitive asteroids.

Samples from asteroid Bennu show that salty, alkaline fluids on its parent body slowly destroyed even the most refractory presolar stardust grains while leaving organic molecules intact, revealing how watery activity can erase a rock’s record of its stellar origins.

Introduction

NASA’s OSIRIS-REx (Origins, Spectral Interpretation, Resource Identification, and Security–Regolith Explorer) mission returned 121.6 g of pristine material from the carbonaceous asteroid (101955) Bennu¹, a rubble-pile body formed by the catastrophic disruption of a larger planetesimal. Remote sensing indicates a hydrated surface with carbonate-filled veins, evidence of extensive aqueous alteration within the parent body^2–5. Returned samples can be grouped into at least three primary morphologies—angular, hummocky, and mottled—based on particle structure, reflectance, mineralogy, and petrology^1,6, consistent with boulder morphologies observed on Bennu’s surface^7–9.

All sample morphologies are dominated by secondary minerals (phyllosilicates, magnetite, carbonates, sulfides, and phosphates) formed during parent-body aqueous alteration^6,10,11. Mineralogical evidence indicates multiple episodes of low-temperature ( < 100 °C) alteration, including formation of evaporite minerals from late-stage alkaline brines^10–12. Some angular and most hummocky particles preserve pre-accretionary components—anhydrous minerals, organic matter, and presolar grains—recording early Solar System processes^1,13,14. Distinct organic chemistries among lithologies are consistent with heterogeneous aqueous reactions in alkaline environments¹⁵, and presolar grains are present across the hummocky and angular materials¹⁴. These nanometer- to micrometer-scale condensates (e.g., nanodiamond, SiC, graphite, oxides, silicates) formed in circumstellar environments or stellar ejecta prior to Solar System formation and are preserved in primitive materials^16–19. In hummocky particles, some clasts show limited aqueous alteration and elevated abundances of O-rich presolar grains, suggesting preservation of protolith material¹⁴.

Mottled particles are less abundant and are associated with bright, carbonate-veined boulders^1,3. They mineralogically resemble angular samples but contain higher abundances of carbonates and Mg- and Na-bearing phosphates^6,10. Mottled particles also exhibit reduced amino acid diversity¹⁵ and lower bulk C and N abundances, largely derived from organic matter²⁰, relative to other lithologies. Here, we characterize the abundance and isotopic composition of presolar grains and isotopically anomalous organic matter within mottled particles and assess their modification during late-stage, low-temperature, alkaline alteration.

Results

Mottled particles – chemical composition and mineralogy

Using digital optical microscopy, scanning electron microscopy (SEM), and energy-dispersive X-ray spectroscopy (EDS), we characterized the mineralogy and petrology of three samples classified as mottled based on their heterogeneous optical appearance and fine-scale textural variability: OREX-800023-16, a polished section from a centimeter-scale stone; OREX-803150-0, a polished section of a millimeter-sized particle; and OREX-803216-0, seven particles embedded in epoxy and polished (Fig. 1 and Methods). In total, these nine mottled particles exhibit a range of shapes, with some displaying irregular outer surfaces reminiscent of hummocky morphology (e.g., OREX-800023-16 and OREX-803216-0 particles (e) and (f)), while others exhibit more angular outlines, such as OREX-803150-0 (Fig. 1).

Fig. 1. Optical reflected-light and backscattered-electron (BSE) images of Bennu samples analyzed in this study.

Optical reflected-light and BSE images of polished section OREX-800023-16 (top left box), thin section OREX-803150-0 (top right box), and aluminum mount OREX-803216-0 (lower box), which contains seven polished Bennu particles investigated in this study. Panels a-g show BSE images of the seven polished Bennu particles.

Most particles contain Mg-, Na-bearing phosphate that fills veins and voids or occurs as coatings on particle surfaces. The only exception is particle (d) of the multi-particle sample OREX-803216-0; this particle is distinguished by a higher abundance of fibrous Fe-sulfides and shows no evidence of Mg- or Na-bearing phosphate on the exterior or in the polished section. Because the mineralogy and petrography of particle (d) differ substantially from those of the other mottled particles—suggesting that its original classification as mottled is not supported by the data—it was excluded from our presolar grain search.

Despite the late-stage phosphate mineralization, the particles are otherwise broadly similar to the angular and hummocky types in their primary matrix mineralogy. They consist of a fine-grained, phyllosilicate-dominated matrix with abundant magnetite framboids, Fe-sulfide grains, and coarser-grained phyllosilicates. In contrast to the hummocky samples studied by Nguyen et al¹⁴., however, most of these mottled particles lack clear clasts or identifiable anhydrous silicates. An exception is OREX-803216-0 particle (f), which shows apparent clastic textures (Fig. 1); however, backscattered electron (BSE) imaging indicates that these textures arise from variations in the distribution of fine- and coarse-grained phyllosilicates.

Presolar Grains – Isotopic Compositions and Abundances

Although we did not find any presolar O-rich grains (silicate or oxide), we identified 27 C-rich grains (median size 208 nm; range 120–415 nm; Table 1) in sections OREX-800023-16 and OREX-803150-0, as well as particles I, (f), and (g) in OREX-803216-0. Figure 2 shows C isotopic maps of two C-rich grains. Of these, 12 are ¹³C-rich, and three are ¹³C-poor (Fig. 3 and Supplementary Data 1). The total abundance is $8_{-1}^{+2}$ ppm (1σ; Fig. 4; Table 1), calculated as the ratio of the cumulative grain area to the total analyzed area²¹, with consistent values across particles ($9_{-3}^{+4}$ ppm in OREX-803150-0; ${12}_{-6}^{+9}$ ppm in OREX-803216-0 (f)). Abundances are uncorrected for detection efficiency to enable direct comparison with the abundances reported for hummocky and angular Bennu samples, Ryugu samples, and carbonaceous chondrites, which were also not corrected. Uncertainties are based on Poisson statistics with small-number confidence intervals²².

Table 1.

Abundances of C-rich and O-rich presolar grains in the different types of Bennu samples

	Mottled	Angular	Hummocky matrix	Clasts in Hummocky	Best Bulk Bennu (B3)
# O-rich grains	0	1	6	18	14
# C-rich grains	27	31	20	9	125
Total area mapped for C & N isotopes (µm2)	58,300	50,736	43,772	16,772	213,180
Total area mapped for O isotopes (µm2)	53,800	50,736	43,853	16,772	208,061
Abundance of C-rich grains (SiC)	$8_{-1}^{+2}$	${55}_{-10}^{+12}$	${38}_{-8}^{+11}$	${33}_{-11}^{+15}$	41 ± 4
Abundance of O-rich grains	~1	$2_{-1}^{+4}$	$7_{-3}^{+4}$	${87}_{-20}^{+26}$	4 ± 1
C-rich Grain Sizes (µm)
Median	0.208	0.330	0.299	0.271	0.265
Maximum	0.421	0.522	0.537	0.364	0.537
Minimum	0.120	0.249	0.216	0.227	0.120

Notes. Best Bulk Bennu (B³) includes data from the mottled material analyzed in this study and from the hummocky and angular morphologies and unsorted aggregate samples^13,14. It excludes data from clasts in hummocky samples. Uncertainties on abundances are 1σ.

Fig. 2. Examples of presolar SiC grains identified in Bennu mottled samples.

False-color ¹³C/¹²C images (top row) showing two ¹³C-rich silicon carbide (SiC) grains identified in mottled samples, along with their corresponding false-color ¹⁴N/¹⁵N images (bottom row). Both grains were found in OREX-803150-0 (Orex-803150-0_122624_12 and OREX-803150-0_100324_32).

Fig. 3. Carbon and nitrogen isotopic compositions of presolar SiC grains in mottled Bennu samples.

The plot shows the ¹²C/¹³C vs ¹⁴N/¹⁵N ratios of twenty-seven presolar silicon carbide (SiC) grains identified in mottled samples, compared with those of SiC grains previously identified in angular and hummocky Bennu materials^13,14, and with SiC grains from different stellar populations⁶². All errors for the isotopic compositions of SiC grains identified in mottled particles can be found in Supplementary Data 1.

Fig. 4. Presolar SiC abundances in the different Bennu sample types and other primitive Solar System materials.

Presolar silicon carbide (SiC) grain abundances in the Bennu mottled samples compared with those in other Bennu sample types¹⁴, Best Bulk Bennu (B³), Ryugu samples^25,26, and carbonaceous chondrites^{24,26,48,63–66}. Error bars represent 1σ uncertainties.

Combining these data with previous measurements of aggregate samples¹³ (unsorted bulk material) and hummocky and angular samples¹⁴, we estimate bulk Bennu abundances (“Best Bulk Bennu”²³ or B³) of 4  ±  1 ppm for O-rich and 41  ±  4 ppm (Table 1) for C-rich presolar grains. The presolar grains from the clasts in the hummocky materials were excluded from these estimates because the clasts are interpreted as representing the initially accreted materials that formed the protolith¹⁴. Including data from the clasts would not significantly change the SiC abundance ( ~ 44 ppm) but would increase the O-rich abundance to 11  ±  2 ppm.

Although not weighted by lithology proportions, these estimates provide a first-order basis for comparison with average presolar grain abundances reported in other astromaterials^18,24–26.

Organic matter (OM) – isotopic compositions and abundances

We identified 537 regions of interest (ROIs) with anomalous isotopic compositions (δ¹⁵N = –838 to 2,674‰; δ¹³C = –229 to 281‰; Fig. 5; Supplementary Data 1), ranging from 196 nm to ~4.7 µm (median 525 nm). Most occur as discrete submicrometer carbon nanoglobules (Fig. 6a), similar to those in Ryugu samples^27,28, whereas others occur in veins or as larger, more diffuse regions (Fig. 6b, c).

Fig. 5. Isotopic compositions of anomalous organic matter in Bennu mottled samples.

Carbon and nitrogen isotopic compositions (δ¹³C vs δ¹⁵N; in permil) of ¹⁵N-rich carbonaceous matter identified in mottled samples compared with those of organic matter from other Bennu samples¹³ and Ryugu samples²⁷. All errors for the isotopic compositions of the organic matter identified in mottled particles can be found in Supplementary Data 2.

Fig. 6. Examples of isotopically anomalous organic matter in Bennu mottled samples.

Secondary electron (SE) images (top row) of ¹⁵N-rich carbonaceous matter, including nanoglobules (red arrows; middle row, (a) and (b)), with their corresponding false-color images of ¹⁴N/¹⁵N (middle row) and ¹³C/¹²C (bottom row), as well as a larger, more diffuse organic matter (OM) region is shown in (c). The ¹³C/¹²C image in (a) also shows a presolar SiC grain (indicated by the black arrow). The images are from OREX-803150-0 (OREX-803150-0_011025_17, OREX803150-0_122324_17, and OREX-803150-0_122624_8).

The abundance of N-anomalous ROIs is 1,886  ±  81 ppm (1σ), based on the areal proportion of the total ROI area to the total mapped area. This is a lower limit as ROIs were identified manually in each image, and not all N-isotopically anomalous ROIs could be reported in every image (e.g., some were too small to define a clear ROI). Additionally, a large carbon-rich area (approximately 1 × 1.5 µm) composed of an accumulation of small nanoglobule-like particles (ranging from a few 100 s nm to ~1 µm in diameter) was identified in OREX-800023-16 (Supplementary Fig. 1). Although some grains are N-anomalous, most have isotopically normal C and N compositions (Supplementary Fig. 2).

Discussion

Earlier studies of Bennu samples reported low abundances of presolar silicate and oxide grains ( ~ 4 ppm)^13,14, consistent with their absence in the mottled particles analyzed here (Fig. 7). These values are comparable to those measured in the hydrated lithology of Ryugu ( ~ 3 ppm)^25,26 and to those observed in petrologic type 2 ungrouped carbonaceous chondrites such as Tagish Lake and Tarda ( ~ 3–4 ppm; Fig. 7)^29,30. In contrast, presolar silicates have not been identified in Bennu angular particles, and only a single oxide grain has been reported ( ~ 2 ppm)¹⁴. Similarly, CI chondrites contain extremely low abundances of O-rich presolar grains, with only one grain identified in Orgueil³¹ and an upper limit of ~4 ppm estimated for Ivuna²⁶, consistent with angular samples from Bennu (Fig. 7).

Fig. 7. Comparison of O- and C-rich presolar grain abundances in Bennu, Ryugu, and aqueously altered carbonaceous chondrites.

a Abundances of O-rich (presolar silicate and oxide) and C-rich (presolar silicon carbide, SiC) grains in mottled Bennu samples compared with other Bennu sample types, Best Bulk Bennu (B³), Ryugu samples^25,26, and carbonaceous chondrites^{24,26,29,30,48,49,63–67}. b Magnified view of the region enclosed by the dashed rectangle in panel (a), showing the distribution of samples with low abundances of O- and/or C-rich presolar grains. Error bars denote 1σ uncertainties.

The absence of O-rich presolar grains in the mottled particles implies an upper abundance limit of ~1 ppm, based on detection limits and typical grain sizes^14,22. Although O-rich grain abundances across Bennu lithologies are broadly similar within uncertainties, a systematic trend emerges when petrographic and mineralogical observations are considered^10,11. First, S-rich clasts within hummocky particles have the highest presolar silicate abundances identified to date, similar to some primitive carbonaceous chondrites¹⁸, and may thus represent Bennu’s protolith. Second, the lower presolar silicate abundance in bulk hummocky particles relative to the S-rich clasts suggests that this material experienced aqueous alteration akin to petrologic type 2 ungrouped carbonaceous chondrites (e.g., Tagish Lake and Tarda), but less than in some type 1 chondrites (Ivuna or Orgueil), where no presolar silicates have been found so far²⁶. Third, the angular particles studied to date contain no remaining presolar silicate grains and only one presolar oxide grain, consistent with a higher degree of aqueous alteration than in the hummocky samples. Finally, the mottled material represents the most extensively altered endmember, lacking all detectable O-rich presolar grains. However, its characteristics indicate not simply a continuation of the same alteration pathway, but rather a chemically distinct late-stage fluid episode superimposed on earlier alteration.

The carbon-rich presolar grain population in the mottled particles provides further insight. Given the low abundance of presolar graphite in Bennu samples ( ~ 1–2 ppm)¹⁴, the 27 C-rich grains identified here are likely SiC. Based on their C and N isotopic compositions, we classify the 27 grains into presolar SiC types and constrain their stellar origins¹⁷ (Fig. 3). Similar to most carbonaceous chondrites, Ryugu samples, and Bennu angular and hummocky particles^14,17,25,26, most of the C-rich presolar grains identified in mottled Bennu particles (22 of 27; 81.5%) have C and N isotopic compositions consistent with mainstream SiC grains, and a few (three of 27; 11%) have compositions consistent with type Z and Y grains (Fig. 3). These 22 grains originated from the C-rich outflows of Asymptotic Giant Branch (AGB) stars^17,32. One grain is ¹³C-rich and exhibits a higher ¹⁵N enrichment than the other grains, consistent with Type X SiC grains, which are thought to have condensed in the ejecta of Type II core-collapse supernovae (SNe)^17,32. The last grain is ¹⁵N-rich and has an extremely low ¹²C/¹³C ratio, consistent with AB SiC grains. The isotopic compositions of ¹⁵N-rich AB SiC grains (also referred to as AB1 grains) are also consistent with an origin from the ejecta of Type II core-collapse SNe^17,33. Compared to angular and hummocky Bennu particles, which show an elevated fraction of supernova-derived SiC ( ~ 31%)¹⁴, we observed a lower proportion ( ~ 7%) in the mottled particles we examined, which is consistent with the one in most carbonaceous chondrites^17,32.

The organic matter (OM) in the mottled particles exhibits a similar range of N and C isotopic compositions (δ¹⁵N between –838 and 2,674‰, and δ¹³C between –229 and 281‰; Fig. 5) to that observed in Bennu aggregate samples (δ¹⁵N between –558 and 3,545‰, δ¹³C between –326 and 364‰)¹³ and in Ryugu samples^25,27,28. These values are also consistent with OM observed in carbonaceous chondrites^27,34–38. Despite the reduced amino acid diversity reported for these samples¹⁵, our data show that the mottled particles host isotopically anomalous OM that is, within uncertainty, indistinguishable from that found in hummocky and angular Bennu samples, Ryugu samples, and carbonaceous chondrites.

This observation stands in sharp contrast with the behavior of presolar grains in the same samples: while presolar SiC was progressively destroyed during late-stage alteration — decreasing from approximately 36–55 ppm in the hummocky and angular lithologies to around 8 ppm in the mottled particles — isotopically anomalous OM was preserved without measurable isotopic resetting. This decoupling demonstrates that OM in carbonaceous asteroids can survive and retain its isotopic signatures under fluid conditions sufficiently destructive to erode highly refractory presolar grains. This differential resilience has direct implications for how presolar and organic inventories should be interpreted in aqueously altered astromaterials: the survival of anomalous OM cannot be taken as evidence that presolar minerals were similarly retained, nor can depleted presolar grain abundances be used to infer that organic reservoirs were correspondingly degraded. The mechanisms governing SiC depletion are therefore distinct from those governing OM preservation and are examined further below.

This resilience may be rooted in the origin and molecular architecture of the OM itself. The origin of OM in astromaterials remains debated, with proposed pathways including partial survival of presolar molecules, low-temperature ion–molecule reactions in interstellar clouds or the outer protoplanetary disk, and in-situ formation during parent-body aqueous alteration^35,37–45. Nittler et al.²⁷ proposed that Ryugu’s OM derived from a macromolecular precursor common to carbonaceous chondrites, formed in the early solar nebula and largely homogenized prior to accretion, with only modest subsequent parent-body modification. Sandford and Gainsforth et al.⁴⁶ similarly identified N- and O-rich, polymer-like OM in Bennu particles, inferred to have formed via low-temperature ice reactions before liquid water existed and to have been only partially modified by later aqueous alteration. Together, these studies suggest that the N-isotopically anomalous OM identified in the mottled particles — and exhibiting comparable isotopic signatures across the hummocky and angular morphologies — likely originated before accretion, in the cold outer solar nebula or a preceding molecular cloud, from a macromolecular precursor whose chemical architecture may render it more resistant to aqueous and brine-driven alteration than mineral phases such as SiC.

The reduced abundance and smaller size of SiC grains in the mottled particles ($8_{-1}^{+2}$ ppm; median size 208 nm; Table 1) contrast sharply with their comparable abundances observed in angular particles, hummocky particles, and clasts within hummocky particles ( ~ 36–55 ppm)¹⁴, as well as in Ryugu samples and most primitive carbonaceous chondrites (Fig. 4)^24–26. Unlike presolar silicate and oxide grains, which are susceptible to aqueous alteration, SiC grains are more resistant to secondary processing⁴⁷. Given the absence of O-rich grains, we cannot simply explain the low abundance of SiC grains in the mottled particles by assuming that these samples experienced the most extensive aqueous alteration among all Bennu lithologies, suggesting that an additional process must be involved, i.e., extended exposure to fluid(s) for longer durations and/or reaction time with fluids (kinetics of fluid–rock reactions).

The uniformity of SiC abundances across most carbonaceous chondrites suggests an initially homogeneous reservoir²⁴. Deviations from this baseline are typically attributed to high-temperature thermal processing, as observed in meteorites such as RBT 04133 (with a peak metamorphic temperature of ~440 °C)⁴⁸ and Adelaide⁴⁹ (Fig. 4). However, Bennu’s hydrothermal alteration temperatures are constrained to <100 °C based on mineralogical observations of Bennu samples^10–12, ruling out thermal metamorphism as the dominant mechanism.

The depletion could partly reflect primary heterogeneity in SiC content among accreted precursor materials of different lithologies; however, this appears unlikely because the hummocky and angular samples and the clasts within hummocky materials exhibit similar abundances of SiC grains. It could also reflect sampling limitations, given the relatively small number of grains analyzed, but we mapped a total area comparable to that of the hummocky and angular samples (Table 1) to ensure consistent overall sampling across all main morphologies. The abundance of SiC grains is also consistent across all mottled particles that we analyzed.

Another possible explanation for the SiC depletion is prolonged interaction with oxidizing alkaline brines. Although direct evidence of SiC dissolution products (e.g., corrosion textures or reaction rims) has not yet been identified in the mottled particles, experimental studies demonstrate that oxidizing alkaline fluids can progressively degrade SiC under low-temperature conditions. While SiC grains are highly resistant to chemical corrosion and aqueous alteration, the presence of an alkaline aqueous environment and sufficient time fundamentally alters the potential for reaction. Experimental studies have shown that, in alkaline fluids, prolonged low-temperature alteration under oxidizing conditions can gradually break down SiC grains by forming silicon dioxide (SiO₂) and other byproducts^50,51. The degradation process is primarily an oxidation reaction accelerated by alkalis. While SiC is known for its high resistance to most acids and alkalis at room temperature, the presence of an oxidizing agent (such as dissolved O) and an alkaline medium (e.g., containing Na⁺ or K⁺ ions) facilitates the reaction. Initially, SiC oxidizes, forming a silica (SiO₂) layer on its surface. In a neutral or mildly acidic environment, this SiO₂ layer is dense and acts as a protective barrier, preventing further reaction. In an alkaline environment, hydroxyl ions (OH^–) and alkali ions such as Na⁺ in the fluid react with the protective SiO₂ layer, dissolving it. The dissolution of the protective silica layer exposes the fresh SiC surface to alkaline and oxidizing fluids, allowing oxidation to proceed more rapidly and unhindered^52–56. Although these reactions are slow at low temperatures, sustained fluid activity over geologic timescales within Bennu’s parent body likely provided the environment to facilitate the oxidation of SiC. In this context, the fluid composition — specifically alkalinity and oxidation state — acted as the primary, rate-limiting control on SiC destruction, because it determines whether oxidative dissolution of the protective SiO₂ layer can proceed at all. The duration of fluid exposure then governed the cumulative extent of grain loss once these threshold conditions were met. This view is consistent with the preservation of SiC at initial baseline abundances in the hummocky and angular lithologies, which experienced earlier, chemically distinct fluids over geologically significant durations without measurable SiC depletion.

The mineralogical context of the mottled particles—including Mg- and Na-bearing phosphates and evaporite salts—indicates the presence of alkaline brines¹⁰. Additional evidence for ammonia-rich fluids and evolving fluid compositions on Bennu’s parent body^15,46,57 supports this interpretation. In this framework, the mottled material represents the product of late-stage fluid alteration of previously altered angular and hummocky materials, during which prolonged exposure to alkaline brines led to the gradual destruction of SiC.

Taken together, these observations define a multistage presolar-grain alteration history within Bennu’s parent body (Fig. 8). Primitive S-rich clasts within hummocky particles preserve the highest presolar silicate abundances and are closest to the original Bennu’s protolith^14,18. The bulk hummocky lithology has a lower abundance of presolar silicate grains, akin to those of petrologic type 2 ungrouped carbonaceous chondrites (e.g., Tagish Lake and Tarda), but a few presolar silicate grains remain, suggesting that it is less altered than type 1 CI chondrites, which have no such grains¹⁴. The angular lithology experienced a higher degree of aqueous alteration than the hummocky lithology, as the angular samples analyzed to date exhibit no remaining presolar silicates and just one remaining presolar oxide grain. The SiC grain abundances of these materials — S-rich clasts, bulk hummocky particles, and angular particles — are similar within uncertainties and consistent with those in CI chondrites²⁴, indicating that, although these three materials have experienced different degrees and stages of aqueous alteration, these different fluid interactions did not significantly affect their SiC grains. The mottled material appears to have undergone late-stage aqueous alteration, unlike the other materials returned from Bennu, as no presolar silicate or oxide grains are present¹⁴. Moreover, it has a much lower abundance of presolar SiC grains, consistent with their slow destruction in an alkaline brine under oxidizing conditions. The three morphologies record a progressive, multi-stage alteration history rather than a single continuous pathway. All three share a common early stage of fluid-rock interaction with a primitive protolith, later superimposed by chemically distinct fluid episodes of increasing severity, with mottled particles representing the most extreme end of the spectrum. Russell et al.⁵⁸ reported that solar anhydrous silicates are absent from the mottled sample and the most altered angular samples that they studied, but are heterogeneously preserved in less altered clasts within hummocky material, consistent with progressive destruction during multistage aqueous alteration. Our observations demonstrate that late-stage alkaline brines on carbonaceous asteroid parent bodies can selectively destroy even highly refractory presolar grains while preserving isotopically anomalous organic matter.

Fig. 8. Model for the preservation and destruction of presolar grains during aqueous alteration on Bennu.

Schematic illustrating the relationship between the distribution and abundance of O- and C-rich presolar grains across different types of Bennu material—hummocky, angular, and mottled particles, as well as certain clasts within hummocky particles—as a function of fluid evolution within Bennu’s parent body. The images of the different particle types are based on backscattered-electron images of samples OREX-800088-9 (hummocky), OREX-800055-18 (angular), and OREX-800023-16 (mottled), and the asteroid image is modified from a mosaic of Bennu (credit: NASA/Goddard/University of Arizona).

Methods

Samples

We analyzed three mottled samples: OREX-800023-16 (a polished section prepared from the mottled stone OREX-800023-0), OREX-803150-0 (a polished thin section of a single millimeter-sized particle), and OREX-803216-0 (an aluminum stud with seven particles embedded in epoxy and polished). The OREX-800023-16 polished section was prepared at NASA Johnson Space Center⁵⁹, and the OREX-803150-0 thin-section was prepared at the University of Arizona. The particle from which OREX-803150-0, and the seven particles composing OREX-803216-0 were hand-picked under a Keyence VHX7000 digital optical microscope from the two aggregate samples OREX-803150-0 and OREX-803011-0, respectively. We selected particles that displayed clear mottled features, such as a white coating on their surfaces.

Optical microscope

All three samples were initially imaged using the Keyence VHX7000 digital optical microscope in the Kuiper-Arizona Laboratory for Astromaterials Analysis (K-ALFAA) at the University of Arizona to obtain reflected-light whole-sample maps (both coaxial and full-ring light mosaics) to aid navigation during subsequent instrument operations. The samples were then coated with carbon.

Secondary electron microscope

We examined the samples using the Hitachi TM4000Plus scanning electron microscope (SEM) at K-ALFAA with a 20 keV electron beam. The Hitachi TM4000Plus SEM is equipped with secondary (SE) and backscatter electron (BSE) imaging detectors, along with an Oxford energy-dispersive X-ray spectrometer (EDS). Backscattered electron mosaic images of the samples were collected to identify suitable fine-grained matrix areas for subsequent isotopic analysis (Fig. 1).

High-resolution nanoscale secondary ion mass spectrometer

Isotopically anomalous presolar grains were identified using the CAMECA High-Resolution Nanoscale Secondary Ion Mass Spectrometer (NanoSIMS-HR) at K-ALFAA. We performed raster ion imaging with a focused Cs⁺ primary beam, with a current of approximately 1–1.2 pA and a diameter of approximately 100 nm. Secondary ions of ^12,13C⁻, ^16,17,18O⁻, and ¹²C^14,15N⁻, along with secondary electrons, were collected simultaneously in multi-collection mode. The mass resolving power (MRP) ranged from 9000 to 12,000 for all detectors, according to CAMECA’s definiton⁶⁰.

To remove the carbon coating and implant primary ions, we initially rastered a high beam current ( ~ 150 pA) over 11 × 11 µm² areas. Each measurement then consisted of 10–20 scans of 10×10 µm^² regions within the pre-sputtered region, with dwell times typically 10,000–15,000 µs per pixel. All measurements were run in chain mode. In total, we mapped an area of 58,300 µm² for C and N isotopes and 53,800 µm² for O isotopes.

A terrestrial kerogen standard deposited onto high-purity Au foil was used to tune the instrument and correct for instrumental mass fractionation of C and N isotopes. Oxygen isotopes were normalized to the surrounding matrix, assuming Solar System values (SMOW). The kerogen standard originates from chert of the Warrawoona Group (Australia, ~3.5 billion years old) and has δ¹³C_PDB and δ¹⁵N_air values of −34.3‰ and ~5.5‰, respectively⁶¹. This kerogen contains 64 wt% C and has a C/N wt% ratio of 181.59 and an H/C ratio of 0.3.

All C, O, and N isotope data were processed using the L’image software (developed by Larry Nittler). We first corrected for detector dead time and then aligned all image frames to account for beam or stage drift. All presolar grains and OM regions of interest were manually located and defined. A grain was considered presolar if its isotopic composition deviated from the average of the surrounding material by more than 4σ and if the anomaly was present in at least three consecutive frames.

Author contributions

Conceptualization: P.H.; Methodology: P.H., J.J.B., E.B.; Sample preparation for analysis: P.H., J.J.B., D.H.; Investigation: P.H., E.B., L.S., J.J.B.; Writing–original draft: P.H.; Writing–review & editing: P.H., J.J.B., E.B., L.S., D.H., D.P.G., J.P.D., H.C.C.; Funding Acquisition: J.J.B., P.H., D.S.L.

Peer review

Peer review information

Nature Communications thanks Kuljeet Marhas and Reto Trappitsch for their contribution to the peer review of this work. A peer review file is available.

Funding

This material is based on work supported by NASA Award NNH09ZDA007O and under Contract NNM10AA11C, issued through the New Frontiers Program. This work was also supported by NASA grant 80NSSC23M0179 (PH). We gratefully acknowledge support for K-ALFAA operations at the University of Arizona from the NASA Planetary Science Enabling Facilities (PSEF) program (grant 80NSSC23K0327), the Lunar and Planetary Laboratory, the College of Science, and the Office for Research, Innovation, and Impact. We also thank the NASA Planetary Major Equipment Program (NNX12AL47G and NNX15AJ22G), NASA’s Early Career Award (80NSSC20K1087), the NSF Major Research Instrumentation Program (1531243 and 0619599), the Gordon and Betty Moore Foundation (grant 9514), the State of Arizona Technology and Research Initiative Fund, and an anonymous donor for supporting instrumentation in K-ALFAA (T.J.Z., J.J.B., P.H., and D.S.L.).

Data availability

All instrument data products underlying the findings of this study are available through the Astromaterials Data System (astromat.org), with the DOIs listed in Supplementary Data 3. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76821-9.