## Main

Between June 2018 and November 2019, the Japan Aerospace Exploration Agency (JAXA) Hayabusa2 spacecraft undertook extensive remote sensing investigations of asteroid Ryugu. Data from the Near-IR Spectrometer (NIRS3) on board Hayabusa2 indicated that Ryugu might be composed of material similar to thermally- and/or shock-metamorphosed carbonaceous chondrite meteorites1; the closest match being the CY (Yamato-type) chondrites2. The low albedo of Ryugu can be explained by the presence of abundant carbon-rich components, in combination with grain size, porosity and space-weathering effects3. The Hayabusa2 spacecraft conducted two touchdowns and sample collections on Ryugu. The first touchdown on 21 February 2019 obtained surface material that was stored in Chamber A of the return capsule, and the second, on 11 July 2019, collected material close to an artificial crater formed by a small carry-on impactor. These samples were stored in Chamber C. Initial, non-destructive Phase 1 characterization of the particles in a dedicated, contamination-free and pure-nitrogen-filled chamber at the JAXA curation facility revealed that Ryugu particles were most similar to CI chondrites4 and displayed ‘a variety of alteration levels’3. The seemingly contradictory classifications of Ryugu, as similar to either CY or CI chondrites, can only be resolved by a detailed isotopic, elemental and mineralogical characterization of Ryugu particles. The results presented here provide a firm basis for determining which of these two preliminary interpretations of asteroid Ryugu’s bulk composition is most likely.

Eight Ryugu particles (~60 mg in total), four from Chamber A and four from Chamber C, were allocated to the Phase 2 curation Kochi team. The principal goals of the research were to elucidate the nature, origin and evolutionary history of the asteroid Ryugu, and to document the similarities to, and differences from, other known extraterrestrial samples such as chondritic meteorites, interplanetary dust particles (IDPs) and returned cometary samples collected by the NASA Stardust mission.

Detailed mineralogical analysis of five Ryugu particles (A0029, A0037, C0009, C0014 and C0068) shows that they consist principally of fine- and coarse-grained phyllosilicates (~64–88 vol%; Fig. 1a,b, Supplementary Fig. 1 and Supplementary Table 1). Coarse-grained phyllosilicates as feathery aggregates (up to several tens of micrometres in size) occur in the fine-grained phyllosilicate-rich matrix (smaller than a few micrometres in size). The phyllosilicate grains are a serpentine–saponite intergrowth (Fig. 1c). A (Si + Al)–Mg–Fe plot further shows that bulk phyllosilicate matrices have intermediate compositions between serpentine and saponite (Fig. 2a,b). Carbonate minerals (~2–21 vol%), sulfide minerals (~2.4–5.5 vol%) and magnetite (~3.6–6.8 vol%) occur in the phyllosilicate matrix. One of the particles examined in this study (C0009) contains a small amount (~0.5 vol%) of anhydrous silicates (olivines and pyroxenes), which can potentially shed light on the original materials that constituted the protolith of Ryugu5. Such anhydrous silicates are rare in the Ryugu particles and have only been positively identified in particle C0009. Carbonates, appearing as fragments in the matrix (smaller than several hundred micrometres), are mainly dolomite, with minor Ca carbonate and breunnerite. Magnetite occurs as isolated grains, framboids, plaquettes or spherical aggregates. Sulfides are mostly pyrrhotite, showing irregular hexagonal prism/plate or lath morphologies. Abundant submicrometre-sized pentlandite occurs in the matrix or in association with pyrrhotite. Carbon-rich phases (<10 µm in size) occur ubiquitously in the phyllosilicate-rich matrix. Other accessory minerals are summarized in Supplementary Table 1. Mineral inventories identified by X-ray diffraction patterns of C0087 and a mixture of A0029 and A0037 are in good agreement with those identified in CI chondrites (Orgueil), but completely different from the CY and CM (Mighei-type) chondrites (Extended Data Fig. 1 and Supplementary Fig. 2). The bulk elemental abundances of Ryugu particles (A0098, C0068) are also consistent with those of CI chondrites6 (Extended Data Fig. 2 and Supplementary Table 2). In contrast, CM chondrites are characterized by depletion of moderate- to highly volatile elements, particularly Mn and Zn, and have higher abundances of refractory elements7. Some elements show highly variable concentrations, which may be a reflection of inherent sample heterogeneity due to the small size of the individual particles and consequential sampling biases. All petrological, mineralogical and elemental characteristics indicate that the Ryugu particles are very similar to CI chondrites8,9,10. The remarkable exception is the absence of ferrihydrite and sulfate in the Ryugu particles, which indicates that these minerals in CI chondrites formed due to terrestrial weathering11.

Bulk oxygen-isotope analysis by laser fluorination (Methods) was undertaken on a 1.83 mg sample of material extracted from particle C0014. For comparison purposes, we ran seven aliquots of Orgueil (CI) (total mass = 8.96 mg) and seven aliquots of Y-82162 (CY) (total mass = 5.11 mg) (Supplementary Table 3).

In Fig. 2d there is a clear separation in terms of both Δ17O and δ18O between the weighted averages for both Orgueil and the Ryugu particles compared with Y-82162. Ryugu particle C0014-4 has a higher Δ17O than Orgueil, although there is overlap at the 2 s.d. level. The higher Δ17O value of the Ryugu particle compared with Orgueil may reflect terrestrial contamination of the latter since its fall in 1864. Weathering in the terrestrial environment11 would necessarily result in the incorporation of atmospheric oxygen and so pull the bulk analysis closer to the terrestrial fractionation line (TFL). This conclusion is consistent with the mineralogical evidence (discussed earlier) that Ryugu particles do not contain ferrihydrite or sulfate, whereas Orgueil does.

These results validate the link between Ryugu particles and CI chondrites based on the mineralogical evidence presented above, while seeming to exclude the possibility of a connection with the CY chondrites. The fact that the Ryugu particles are not related to the CY chondrites, which show clear mineralogical evidence for dehydration, is puzzling. Orbital observations of Ryugu seemed to indicate that it had experienced a dehydration and so was probably composed of CY material2. The reason for this apparent discrepancy remains unclear. Oxygen-isotope analysis of other Ryugu particles is presented in a companion paper12. However, the results from this extended dataset are also consistent with a link between Ryugu particles and CI chondrites.

Using coordinated microanalysis techniques (Supplementary Fig. 3), we studied the spatial distribution of organic carbon throughout the entire surface area of the C0068,25 focused-ion-beam (FIB) section (Fig. 3a–f). Carbon near-edge X-ray absorption fine-structure (NEXAFS) spectra in the C0068,25 section show a variety of functional groups—aromatic or C=C (285.2 eV), C=O (286.5 eV), C–H (287.5 eV) and C(=O)O (288.8 eV)—without the graphene structure at 291.7 eV (Fig. 3a), implying low degrees of thermal alteration. The strong C–H peak (287.5 eV) of the organics in the C0068,25 section is distinct from the previously studied insoluble organic matter of carbonaceous chondrites13 and shows more similarities with IDPs14 and cometary particles obtained by the Stardust mission15. The strong C–H peak at 287.5 eV and the very weak aromatic or C=C peak at 285.2 eV imply that the organics are aliphatic-rich (Fig. 3a and Supplementary Fig. 3a). The aliphatic-rich organic areas are present locally within coarse-grained phyllosilicates, as well as within the regions with a poorly aromatic (or C=C) carbon structure (Fig. 3c,d). In contrast, the A0037,22 (Supplementary Fig. 3) section displays a lower abundance of aliphatic carbon-rich areas. The bulk mineralogy of these particles is carbonate-rich, similar to CI chondrites16, and this is indicative of extensive parent body aqueous alteration (Supplementary Table 1). Oxidizing conditions would promote higher concentrations of carbonyl and carboxylic functional groups in the organic matter association with carbonates17. The submicrometre-scale distribution of organics with an aliphatic carbon structure may vary strongly depending on the distribution of coarse-grained phyllosilicates. A hint of aliphatic-bearing organics in association with phyllosilicate-OH has been reported in the Tagish Lake meteorite18. The coordinated microanalysis data suggest that aliphatic-rich organics may be widely distributed in C-type asteroids and exist in close association with phyllosilicates. This inference is consistent with a previous report of aliphatic/aromatic CH in the Ryugu particles demonstrated by MicrOmega, a hyperspectral microscope operating in the near-infrared range3. An important and as-yet-unresolved question is whether the unique nature of the aliphatic carbon-rich organics associated with coarse-grained phyllosilicates observed in this study is found only in the Ryugu asteroid.

A kinetic study of organic matter degradation in the Murchison meteorite19 may provide important insight into the heterogeneous distribution of aliphatic-rich organics seen in the Ryugu particles. This study suggests that the aliphatic C–H bonds in organics survive at the maximum temperature of ~30°C on the parent body and/or will be lost depending on a time–temperature relationship (that is, 200 years at 100 °C and 100 million years at 0 °C). The initial distribution of aliphatic-rich organics in phyllosilicates could be preserved if the parent body was not heated beyond the time limit at a given temperature. However, aqueous alteration on the parent body might complicate this interpretation because carbonate-rich A0037 does not show any aliphatic carbon-rich regions associated with phyllosilicates. This low alteration temperature is broadly consistent with the occurrence of cubanite in Ryugu particles (Supplementary Table 1)20.

A large nanoglobule was found in the C0068,25 section (n.g-1; Fig. 3a–c,e) showing highly aromatic (or C=C), moderately aliphatic and weakly C(=O)O and C=O spectra. The aliphatic carbon feature does not match that of the bulk insoluble organic matter and organic nanoglobules associated with chondritic meteorites (Fig. 3a)17,21. Raman and infrared spectroscopy analyses of nanoglobules in Tagish Lake show that they are composed of aliphatic and oxidized organic matter and disordered polycyclic aromatic organic matter, all of which are structurally complex22,23. The aliphatic carbon feature seen in n.g-1 may be an analytical artefact due to the surrounding matrix containing aliphatic-rich organics. Interestingly, n.g.-1 contains an embedded amorphous silicate (Fig. 3e), which is a texture that has not been reported for any extraterrestrial organics so far. The amorphous silicate could be a pristine component indigenous to n.g-1 or have originated from amorphization of hydrous/anhydrous silicate by ion- and/or electron-beam irradiation during analysis.

NanoSIMS ion images (Fig. 3f) of the C0068,25 section display homogeneous variation in δ13C and δ15N, except for a presolar grain (P.G-1 in the δ13C image in Fig. 3f) with a large 13C enrichment of 30,811‰ (Supplementary Table 4). X-ray elemental and high-resolution TEM images of the grain show only the carbon concentration and the basal plane spacing of 0.3 nm, corresponding to graphite. It is worth noting that δD (841 ± 394‰) and δ15N (169 ± 95‰) values of aliphatic-rich organics associated with coarse-grained phyllosilicates seem to be slightly higher than the average for the entire C regions (δD = 528 ± 139‰, δ15N = 67 ± 15‰) in C0068,25 (Supplementary Table 4). This observation indicates that aliphatic-rich organics in coarse-grained phyllosilicates could be more primitive than the surrounding organics because the latter may have undergone isotopic exchange with surrounding water in the parent body24. Alternatively, these isotopic variations could also be linked to the initial formation process. In CI chondrites, the fine-grained phyllosilicates are interpreted to have formed via continued alteration of the initial coarse-grained clusters of anhydrous silicates16. Aliphatic-rich organics might have formed from precursor molecules either in the protoplanetary disk, or interstellar medium before the Solar System formed24 and were then slightly modified during aqueous alteration in the Ryugu (grand)parent body. The size (<1.0 km) of Ryugu is too small to sufficiently maintain internal heat for aqueous alteration to form hydrous minerals25. Therefore, a precursor body to Ryugu of a few tens of kliometres in size may be required. Aliphatic-rich organics may have maintained their original isotopic ratios due to their association with coarse-grained phyllosilicates. However, the exact nature of the isotopically heavy carrier is still uncertain due to a complex fine-scale mixture of various components in these FIB sections. It could be either the aliphatic-rich organics or the surrounding coarse-grained phyllosilicates in Ryugu particles. Note that the organic matter is generally more enriched in D than the phyllosilicates in almost all carbonaceous chondrites, including CI chondrites, with the exception of the CM Paris meteorite24,26.

A bulk δD and δ15N plot of the A0002,23 and A0002,26, A0037,22 and A0037,23 and C0068,23, C0068,25 and C0068,26 FIB sections (a total of seven FIB sections from three Ryugu particles) obtained by NanoSIMS is shown in Fig. 4 (Supplementary Table 4) in comparison with other Solar System objects27,28. Variations in bulk δD and δ15N in the A0002, A0037 and C0068 sections are compatible with those seen in IDPs but are higher than CM and CI chondrites (Fig. 4). Note that the range of δD values (−240 to 1,655‰) in Stardust cometary samples29 is larger than that of Ryugu. The bulk δD and δ15N of the Ryugu sections tend to be lighter than the average values of Jupiter family and Oort cloud comets (Fig. 4). The lower δD values for CI chondrites may reflect the influence of terrestrial contamination in these samples30. Given the similarities among Bells, Tagish Lake and IDPs, the large heterogeneity of the δD and δ15N values in Ryugu particles may represent variations of the initial isotopic signatures of their organic and water components in the early Solar System26. The similar isotopic variations in δD and δ15N in Ryugu particles and IDPs indicate that both could have formed from materials that came from the same source. IDPs are thought to have been derived from cometary sources14. Therefore, Ryugu could contain cometary-like and/or at least outer Solar System materials. However, it may be more complex than we claim here because of (1) a mixture of chondritic water and D-rich water on the parent body31 and (2) the D/H ratio of comets depending on cometary activities32. However, the causes of the hydrogen and nitrogen isotopic heterogeneities observed in the Ryugu particles are not yet fully understood, due in part to the limited numbers of analyses available thus far. The results from hydrogen and nitrogen isotopic systems still raise the possibility that Ryugu contains a substantial portion of outer Solar System-derived materials and so could show some similarities with comets. The Ryugu sections show no obvious correlation between δ13C and δ15N (Supplementary Table 4).

The delivery of volatiles (that is, organics and water) to the Earth is still a subject of notable debate26,27,33. The submicrometre-scale organics associated with coarse-grained phyllosilicates in Ryugu particles, identified in this study, probably represent one important source of volatiles. Organics incorporated into coarse-grained phyllosilicates seem to be better protected from degradation16,34 and breakdown35 than those within fine-grained matrix. The heavier hydrogen isotopic composition of the particles means that they are unlikely to be the only source of volatiles delivered to the early Earth. They may have been mixed with components with lighter hydrogen isotopic compositions, as recently proposed by the hypothesis of solar-wind-derived water in silicates36.

In this study we demonstrate that CI meteorites, despite their geochemical importance as proxies of the bulk Solar System composition6,10, are terrestrially contaminated samples. We also provide direct evidence of an interaction between aliphatic-rich organics and adjacent hydrous minerals and show that Ryugu could contain outer Solar System-derived materials37. The findings of this study clearly demonstrate the importance of direct sampling of primitive asteroids and the need to transport returned samples in totally inert and sterile conditions. The evidence presented here shows that Ryugu particles are undoubtedly among the most uncontaminated Solar System materials available for laboratory study and ongoing investigations of these precious samples will certainly expand our understanding of early Solar System processes. Ryugu particles are the best proxy we have for the bulk composition of the Solar System.

## Methods

### Designs of analyses of Ryugu particles

To determine the complex microtexture and chemical characteristics of the sample at submicrometre scales, we conducted coordinated microanalysis utilizing synchrotron radiation-based computed tomography (SR-XCT) and SR-X-ray diffractometry (XRD)-CT, FIB–STXM-NEXAFS–NanoSIMS–TEM analysis without degradation, contamination due to the terrestrial atmosphere and small particles or mechanical sample damage. In parallel, we carried out systematic bulk analysis with scanning electron microscopy (SEM)-EDS, EPMA, XRD, instrumental neutron activation analysis (INAA) and a laser fluorination oxygen-isotope facility. The analytical procedure is shown in Supplementary Fig. 3, and each analysis is described in the following sections.

### Sample transportation and handling processes

The Ryugu asteroid particles were recovered from the Hayabusa2 re-entry capsule and transported to the JAXA Curation Facility at Sagamihara, Japan, without terrestrial atmospheric contamination4. After initial and non-destructive characterizations at the JAXA Curation Facility, an airtight facility-to-facility transfer container and a sample capsule pack (made of sapphire glass and stainless steel with 10 or 15 mm diameter depending on sample size) were used to avoid chemical reactions with the surrounding environment and/or terrestrial contaminants (for example, water vapour, hydrocarbons, atmospheric gases and small particles) and cross-contamination between samples during sample preparation and transportation among institutes and universities38. To avoid degradation and contamination due to interaction with the terrestrial atmosphere (water vapour and oxygen gas), all sample preparations (including chipping by a tantalum chisel, and cutting by a counter-balanced diamond wire saw (Meiwa Fosis Corporation DWS 3400) and epoxy mount preparation) were conducted in a glove box in an atmosphere of pure, dry N2 (Dew point: −80 to −60 °C, O2 ~50–100 ppm). All items used here were cleaned by a combination of ultrapure water and ethanol under ultrasonication with different frequencies.

We studied meteorite collections (CI: Orgueil, CM2.4: Yamato (Y)-791198, CY: Y-82162, and CY: Y 980115) of the Antarctic Meteorite Research Center at the National Institute of Polar Research (NIPR) in this study.

For the transfers between instruments for SR-XCT, NanoSIMS, STXM–NEXAFS and TEM analyses we used the universal holders for ultrathin film samples described in previous studies38.

### SR-XCT analysis

The SR-XCT analyses for Ryugu samples were performed with the integrated CT system at BL20XU/SPring-8. The integrated CT system consists of different measurement modes: the wide-field and low-resolution (WL) mode for capturing the entire structure of the sample, the narrow-field and high-resolution (NH) mode for precise measurement of the region of interest and the XRD mode to acquire the diffraction pattern of the bulk volume of the sample and perform XRD-CT to acquire 2D mineral phase maps of horizontal planes in the sample. Note that all measurements could be done without removing the sample holder from the mount using the integrated system, and therefore accurate measurements for CT and XRD-CT were possible. The X-ray detector for the WL mode (BM AA40P; Hamamatsu Photonics) was equipped with a complementary metal oxide semiconductor (CMOS) camera that had 4,608 × 4,608 pixels (C14120-20P; Hamamatsu Photonics), a scintillator consisting of a 10-μm-thick lutetium aluminium garnet (Lu3Al5O12:Ce) single crystal and relay lenses. The pixel size of the WL mode was approximately 0.848 µm. Thus, the field of view (FOV) of the WL mode was ~6 mm with the offset CT scan mode. The X-ray detector for the NH mode (BM AA50; Hamamatsu Photonics) was equipped with a scintillator consisting of gadolinium aluminium gallium garnet (Gd3Al2Ga3O12) that was 20 µm thick, a CMOS camera that had 2,048 × 2,048 pixels (C11440-22CU; Hamamatsu Photonics) and a ×20 lens. The pixel size of the NH mode was ~0.25 µm and the FOV was ~0.5 mm. The detector for XRD mode (BM AA60; Hamamatsu Photonics) was equipped with a scintillator consisting of a P43 (Gd2O2S:Tb) powder screen that was 50 µm thick, a CMOS camera that had 2,304 × 2,304 pixels (C15440-20UP; Hamamatsu Photonics) and relay lenses. The effective pixel size of the detector was 19.05 µm and the FOV was 43.9 mm2. To increase the FOV, we applied the offset CT procedure in the WL mode. Transmitted light images for CT reconstruction were composed by combining images of 180° to 360° that horizontally reflected around the rotation axis, with the images from 0° to 180°.

For the XRD mode, an X-ray beam was focused by a Fresnel zone plate. With this mode, the detector was located 110 mm behind the sample, with a beam stop 3 mm in front of the detector. Diffraction images of 2θ ranging from 1.43° to 18.00° (lattice spacing d = 16.6–1.32 Å) were acquired with an X-ray beam spot focusing on the bottom of the FOV of the detector. Samples were vertically translated with a certain interval, and a half-rotated for each vertical scan step. Diffraction of mineral grains in a horizontal plane can be obtained if the mineral grains meet the Bragg condition in the 180° rotation. Diffraction images were then integrated into one image for each vertical scan step39. The SR-XRD-CT analytical conditions were almost the same as for the SR-XRD analyses. For the XRD-CT mode, the detector was located 69 mm behind the sample. Diffraction images of 2θ ranging from 1.2° to 17.68° (d = 19.73–1.35 Å) were acquired with both an X-ray beam and a beam stop that were placed on a straight line to the centre of the FOV of the detector. The sample was scanned horizontally with 180° rotation of the sample. The SR-XRD-CT image was reconstructed using the peak intensity of minerals as the pixel value. Samples were typically scanned with 500 to 1,000 steps for a horizontal scan.

The X-ray energy was fixed to 30 keV for all experiments because it is the lower limit for X-ray penetration of meteorites that are ~6 mm in diameter40. The number of images acquired for all CT measurements during the 180° rotation was 1,800 (3,600 for the offset CT procedure), and the exposure time for images was 100 ms for the WL mode, 300 ms for the NH mode, 500 ms for XRD and 50 ms for XRD-CT. The typical scanning time for one sample was ~10 min in the WL mode, ~15 min in the NH mode, ~ 3 h for XRD, and 8 h for SR-XRD-CT.

CT images were reconstructed by a convolution-backprojection method and normalized for linear attenuation coefficients of 0 to 80 cm−1. Slice software was applied for the analysis of 3D data, and muXRD software was used for the analysis of XRD data.

### Optical, SEM–EDS and EPMA analyses

The epoxy-mounted Ryugu particles (A0029, A0037, C0009, C0014 and C0068) were polished on a surface in a stepwise manner down to the level of a 0.5 µm diamond lapping film (3M) under dry conditions avoiding elution of any materials from the surface during polishing. A polished surface of each sample was examined first by an optical microscope, and then by a JEOL JSM-7100F SEM equipped with an energy-dispersive spectrometer (AZtec energy) to obtain an overview of the mineralogy and textures of the samples by backscattered electron (BSE) images and qualitative elemental maps at NIPR. Major and minor elemental abundance analysis of each sample were conducted with an electron probe microanalyser (EPMA, JEOL JXA-8200). Phyllosilicate and carbonate grains were analysed with a current of 5 nA, and sulfide, magnetite, olivine and pyroxene with a current of 30 nA using natural and synthetic standards at 15 keV. Modal abundances were calculated from elemental maps and BSE images by utilizing the ImageJ 1.53 software with a proper threshold arbitrarily set for each mineral.

### High-precision oxygen isotopic analysis

Oxygen isotopic analysis was undertaken at the Open University (Milton Keynes, UK) using an infrared laser-assisted fluorination system41. The Hayabusa2 samples were transported to the Open University in nitrogen-filled facility-to-facility transfer container38.

Ryugu particle C0014-4 and the Orgueil (CI) samples were run in modified ‘single shot’ mode42, whereas analyses of Y-82162 (CY) were run in a single tray with multiple sample wells41. Owing to their anhydrous composition, it was not necessary to use the single-shot technique for the CY chondrites. Sample heating in the presence of BrF5 was carried out using a Photon Machines Inc. 50 W infrared CO2 laser (10.6 μm) mounted on an XYZ gantry. The progress of the reaction was monitored by means of an integrated video system. After fluorination, the released O2 was purified by passing it through two cryogenic nitrogen traps and over a bed of heated KBr to remove any excess fluorine. The isotopic composition of the purified oxygen gas was analysed using a Thermo Fisher MAT 253 dual-inlet mass spectrometer with a mass resolving power of approximately 200.

In some cases the amount of O2 gas liberated during sample reaction was less than 140 µg, the approximate limit for using the bellows facility on the MAT 253 mass spectrometer. In these cases, analysis was undertaken using the microvolume. Following analysis of the Hayabusa2 particle, the internal obsidian standard was fluorinated, and its oxygen-isotope composition determined.

The NF+ fragment ion of NF3+ can cause interference with the mass 33 beam (16O17O). To eliminate this potential problem, most samples were treated using a cryogenic separation procedure. This was either done in the forward sense before analysis on the MAT 253 or as a second analysis with the already analysed gas pulled back onto a dedicated molecular sieve and then rerun after cryogenic separation. Cryogenic separation involved taking the gas onto the molecular sieve at liquid nitrogen temperature and then releasing it to the main molecular sieve at a temperature of −130°C. Extensive tests have shown that NF+ is retained on the first molecular sieve and that no substantial fractionation results from the use of this technique.

The overall system precision in bellows mode, as defined by replicate analyses of our internal obsidian standard43, was: ±0.053‰ for δ17O; ±0.095‰ for δ18O; ±0.018‰ for Δ17O (2 s.d.). Oxygen isotopic analyses are reported in standard δ notation, where δ18O has been calculated as:

$$\delta ^{18}{{{\mathrm{O}}}} = \left[ {\left( {{\,}^{18}{{{\mathrm{O}}}}/{\,}^{16}{{{\mathrm{O}}}}} \right)_{{{{\mathrm{sample}}}}}/\left( {{\,}^{18}{{{\mathrm{O}}}}/{\,}^{16}{{{\mathrm{O}}}}} \right)_{{{{\mathrm{VSMOW}}}}} - 1} \right]1,000$$

and similarly for δ17O using the 17O/16O ratio. VSMOW is the international standard Vienna Standard Mean Ocean Water. Δ17O, which represents the deviation from the terrestrial fractionation line, has been calculated as: Δ17O = δ17O − 0.52 × δ18O. A blank correction has been applied to all the data presented in Supplementary Table 3.

### Sample preparation using a FIB

Approximately 150- to 200-nm-thick sections were extracted from Ryugu particles using a Hitachi High Tech SMI4050 FIB instrument at the Kochi Institute for Core Sample Research, JAMSTEC. Note that all FIB sections were extracted from unprocessed broken pieces of original particles just after removal from facility-to-facility transfer containers filled with N2 gas. The pieces were not measured by SR-CT and were handled with minimal exposure to the terrestrial atmosphere to avoid potential damage and contamination that could influence the carbon K-edge spectra. After deposition of tungsten protection layers, regions of interest (up to 25 × 25 µm2) were cut out and thinned using a Ga+ ion beam at an accelerating voltage of 30 kV and then finalized at 5 kV and a probe current of 40 pA to minimize damage to surface layers. The ultrathin sections were subsequently mounted on scaled-up Cu grids (Kochi grids)39 using a micromanipulator equipped with the FIB.

### Elemental abundance measured by INAA

Ryugu particles A0098 (1.6303 mg) and C0068 (0.6483 mg) were doubly sealed in cleaned and high-purity polyethylene sheets in a pure N2-filled glove box at SPring-8 without any interaction with the terrestrial atmosphere. Sample preparation of JB-1 (a geological standard rock sample issued by the Geological Survey of Japan) was carried out at Tokyo Metropolitan University.

INAA was performed at the Institute for Integrated Radiation and Nuclear Science, Kyoto University. Samples were irradiated two times with different irradiation periods chosen according to the half-lives of the nuclides used for elemental quantification. First, samples were irradiated for 30 s in pneumatic irradiation tube no. 3 with thermal and fast neutron fluxes of 4.6 × 1012 and 9.6 × 1011 cm−2 s−1, respectively, to determine Mg, Al, Ca, Ti, V and Mn abundances. Chemical reagents such as MgO (99.99% purity; Soekawa Chemical), Al (99.9% purity; Soekawa Chemical) and Si metals (99.999% purity; FUJIFILM Wako Pure Chemical) were also irradiated to correct for interfering nuclear reactions such as (n,p). Sodium chloride (99.99% purity; MANAC) was also irradiated with the samples to correct for neutron flux variations.

After neutron irradiation, the outer polyethylene sheet was replaced with a new sheet and gamma rays emitted from the samples and reference standards were immediately measured using Ge detectors. The same samples were re-irradiated for 4 h in pneumatic irradiation tube no. 2 with thermal and fast neutron fluxes of 5.6 × 1012 and 1.2 × 1012 cm−2 s−1, respectively, to determine Na, K, Ca, Sc, Cr, Fe, Co, Ni, Zn, Ga, As, Se, Sb, Os, Ir and Au abundances. Reference samples for Ga, As, Se, Sb, Os, Ir and Au were prepared by dropping a proper amount (10 to 50 µg) of concentration-known standard solutions of these elements onto the two sheets of filter paper, which were then irradiated with the samples. Gamma-ray counting was carried out at the Institute for Integrated Radiation and Nuclear Science, Kyoto University and RI Research Center, Tokyo Metropolitan University. The analytical procedure and reference materials for elemental quantification by INAA were the same as those described in our previous work44.

### Bulk mineralogy determined using XRD

An X-ray diffractometer (Rigaku SmartLab) was used to collect the diffraction patterns of Ryugu samples A0029 (<1 mg), A0037 (1 mg) and C0087 (<1 mg) at NIPR. All of the samples were ground to fine powders using a sapphire glass plate on a silicon non-reflection plate, and then homogeneously distributed onto the silicon non-reflection plate without any fluid (water or alcohol). The measurement conditions were as follows: Cu Kα X-rays were produced at 40 kV tube voltage and 40 mA tube current; the length of the limiting slit was 10 mm, the divergence angle was (1/6)°, the in-plane rotation speed was 20 r.p.m., the 2θ range (two-fold Bragg angle) was 3–100°, and it took ~28 h for each analysis. Bragg Brentano optics were used. The detector was a one-dimensional silicon semiconductor detector (D/teX Ultra 250). Cu Kβ X-rays were removed using a Ni filter. The peaks were identified by a comparison with the measured data using the available samples, synthesized Mg-saponite (JCSS-3501, Kunimine Industries CO. Ltd), serpentine (antigorite, Miyazu, Nichika) and pyrrhotite (monoclinic 4C, Chihuahua, Mexico) and using the powder diffraction file data from International Center for Diffraction Data, dolomite (PDF 01-071-1662) and magnetite (PDF 00-019-0629). The diffraction data for Ryugu were also compared with data for hydrously altered carbonaceous chondrites, Orgueil CI, Y-791198 CM2.4 and Y 980115 CY (heating stage III, 500–750 °C). The comparison showed similarities with Orgueil but not with Y-791198 and Y 980115.

### Carbon functional groups identified using STXM–NEXAFS

The carbon K-edge NEXAFS spectra of ultrathin section samples made by FIB were measured using the STXM beam line, BL4U, at the UVSOR Synchrotron Facility, Institute for Molecular Science (Okazaki, Japan). The beam spot size focused with Fresnel zone-plate optics was about 50 nm. The energy step size was 0.1 eV in the fine-structure portions of the near-edge region (283.6–292.0 eV), and 0.5 eV in the pre-edge and post-edge regions (280.0–283.5 eV and 292.5–300.0 eV). The acquisition time per image pixel for each energy step was set to 2 ms. Helium gas at a pressure of ~20 mbar was backfilled with the STXM analysis chamber after evacuation. This helps to minimize the thermal drift of X-ray optics-related apparatus in the chamber and the sample holder, and to reduce sample damage and/or oxidation. The carbon K-edge NEXAFS spectra were obtained from stack data using the aXis2000 software and the in-house customized software for STXM data reduction. Note that an in-house, custom built sample transfer vessel and a glove box were used to avoid oxidization and contamination on the sample.

### Hydrogen, carbon, and nitrogen isotopic imaging analysis using NanoSIMS

The hydrogen, carbon and nitrogen isotopic compositions of the Ryugu FIB sections were analysed using isotopic imaging with the JAMSTEC NanoSIMS 50L after STXM–NEXAFS analysis. A focused primary Cs+ beam of approximately ~2 pA was used for carbon, and nitrogen isotopic analyses, and approximately ~13 pA was used for hydrogen isotopic analyses, rastered over approximately 24 × 24 μm2 to 30 × 30 μm2 areas on the samples. Each analysis was initiated after stabilization of the secondary-ion-beam intensity following 3 min of pre-sputtering with a relatively strong primary-ion-beam current. For carbon and nitrogen isotopic analysis, images of 12C, 13C, 16O, 12C14N and 12C15N were acquired simultaneously by multidetection with seven electron multipliers at a mass resolving power of approximately 9,000, which was sufficient to separate all relevant isobaric interferences (that is, 12C1H on 13C and 13C14N on 12C15N). For hydrogen isotopic analysis, images of 1H, 2D and 12C were acquired using three electron multipliers in multidetection at a mass resolving power of approximately 3,000. Each analysis consisted of 30 scanned images of the same area, with individual images consisting of 256 × 256 pixels for the carbon and nitrogen isotopic analyses and 128 × 128 pixels for the hydrogen isotopic analysis. The dwell times were 3,000 μs per pixel for the carbon and nitrogen isotopic analyses and 5,000 μ per pixel for the hydrogen isotopic analysis. We used 1-hydroxybenzotriazole hydrate as the hydrogen, carbon and nitrogen isotopic standards to correct for instrumental mass fractionation45.

For silicon isotopic compositions of the presolar graphite in the C0068-25 FIB section, we acquired secondary-ion images of 12C, 13C, 16O, 28Si, 29Si and 30Si simultaneously by multidetection with six electron electron multipliers at a mass resolving power of approximately 9,000. These images consist of 256 × 256 pixels with a dwell time of 3,000 µs per pixel. We used a silicon wafer as the hydrogen, carbon and silicon isotopic standard to correct for instrumental mass fractionation.

The isotopic images were processed using the software NASA JSC imaging software for NanoSIMS45. Data were corrected for electron multiplier dead time (44 ns), and the quasi-simultaneous arrival effect. Different scans of each image were aligned to correct image drift during acquisition. Final isotopic images were generated by adding the secondary ions of each image from each pixel over the scans.

### Micrometre- to submicrometre-scale mineralogical observations using TEM

After STXM–NEXAFS and NanoSIMS analysis, the same FIB sections were studied using a transmission electron microscope (JEOL JEM-ARM200F) operated at an accelerating voltage of 200 kV at Kochi, JAMSTEC. Microtextural observations were performed by bright-field TEM and high-angle annular dark-field scanning TEM imaging. Mineral phases were identified using selected-area electron diffraction and lattice-fringe imaging, and chemical analyses using EDS with a 100 mm2 silicon drift detector and JEOL Analysis Station 4.30 software. For quantitative analyses, the intensities of the characteristic X-rays of each element were measured using a fixed acquisition time of 30 s, a beam scan area of ~100 × 100 nm2 and a beam current of 50 pA in scanning TEM mode. The (Si + Al)–Mg–Fe ratios of phyllosilicates were determined using experimental thickness-corrected k factors obtained from a natural pyrope-almandine garnet standard.