Main

The Japan Aerospace Exploration Agency (JAXA)’s Hayabusa2 spacecraft explored the near-Earth C-type asteroid Ryugu and brought back regolith materials1. Ryugu is a rubble-pile asteroid that consists of numerous fragments of an original, larger parent body2. Ryugu samples show a close affinity with CI carbonaceous chondrites, which are the most chemically primitive, the most hydrated and organic-rich groups of asteroidal materials3,4,5,6,7,8. The accretion region of the parent body is probably beyond the H2O and CO2 snow lines (3-4 au) and possibly beyond the orbit of Jupiter4,7,9,10,11, where numerous icy planetesimals could have formed12. Therefore, Ryugu samples offer valuable insights into the evolution and water activity of outer Solar System bodies, including icy bodies that may possess habitable water environments.

Ryugu samples are mainly composed of phyllosilicates and other secondary minerals, indicating extensive aqueous alteration of their parent body3,4,11,13. Sodium compounds are particularly important in the aqueous alteration because the alteration probably proceeded with sodium-rich solutions throughout the relevant period4,8. This is evidenced by the enrichment of sodium ions in water extracted from Ryugu samples14 and by the presence of sodium-rich phyllosilicates and sodium-magnesium phosphates4,8. A previous analysis suggested that there were unidentified sodium phases that could easily have been decomposed on Earth or during sample analysis8. The characterization of unknown sodium-bearing minerals will unveil the progress of chemistry in solutions within the parent body. Ryugu samples have been recovered and stored without direct exposure to the terrestrial atmosphere15, which allows them to be studied in their original state on the asteroid. In this study, we performed a non-destructive observation of sodium-rich Ryugu particles, followed by chemical and crystallographic analysis of the samples at the submicrometre scale.

Results

Optical microscopy, SEM, XCT and XRD analyses

Ryugu grain C0071, which has a diameter of ~1.5 mm (Fig. 1a), was examined using scanning electron microscopy (SEM), synchrotron radiation-based X-ray computed tomography (SR-XCT) and synchrotron radiation-based X-ray diffraction (SR-XRD). Additionally, the assembly of fine grains C0369, which have diameters of less than 1 mm, was examined using SEM. The SEM observations and SR-XRD analysis showed that the grain C0071 is mainly composed of phyllosilicates (saponite and serpentine), iron- and nickel-bearing sulfides, magnetite (framboidal, spherulitic and plaquette shapes) and dolomite as well as lesser amounts of iron-rich magnesite, calcium phosphate and magnesium sodium phosphate (Fig. 1b,c and Supplementary Fig. 1). The phyllosilicate matrix on the surface of an ~0.6 mm2 area is rich in sodium (Extended Data Fig. 1b,d). Its distribution is interrupted at the surface boundaries, with no sodium concentrations in the adjacent surfaces. This suggests that the sodium-rich surface existed before this Ryugu grain acquired its current shape and size through fracturing. Several sodium-rich patches containing chlorine are present on this surface (Supplementary Fig. 2). In addition, the grain surface contains a sodium-rich vein near the edge, which appeared white when observed under reflected light using an optical microscope (Fig. 1a). This vein has a width ranging from 6 to 20 µm and a length extending up to 500 µm (Fig. 1b and Extended Data Fig. 1e). Note that the sodium-rich vein was observed by SEM without being exposed to the atmosphere (Methods), indicating that it is not a product of terrestrial weathering or contamination. The sodium-rich grain surface has rounded and frothy phyllosilicates (Extended Data Fig. 1f). The morphology corresponds to space weathering due to micrometeorite bombardment and solar wind irradiation16.

Fig. 1: Sodium carbonate vein on Ryugu grain C0071.
figure 1

a, Optical microscope image of the Ryugu grain. The sodium carbonate vein is indicated by an arrow. b, Composite elemental map of sodium (green), silicon (blue) and iron (magenta), obtained by SEM. c, Composite elemental map of magnesium (cyan), phosphorus (yellow) and calcium (magenta). Ap, apatite (indicated by dashed white arrows); Do, dolomite; Mgt, magnetite; Na-car, sodium carbonates; Na-Mg-ph, Na-Mg phosphates (indicated by solid yellow arrows); Phy, phyllosilicate matrix.

Full size image

Three fine grains among the ~200 grains in C0369 have sodium-rich veins of width ≲10 µm and length ≲140 µm (Extended Data Table 1 and Extended Data Figs. 2–4). The veins are also near the edge of the grains and show relatively smooth surfaces compared to the surrounding phyllosilicate matrix (Extended Data Fig. 3d). Some show granular morphologies (Extended Data Fig. 2c). The observed fine grains are mainly composed of phyllosilicates, iron- and nickel-bearing sulfides, magnetite and calcium-magnesium-iron carbonates. The grain surfaces adjacent to the sodium-rich veins include frothy phyllosilicates and magnetite with porous surfaces, indicating space weathering16,17 (Extended Data Figs. 2–4).

TEM and STXM analyses

We extracted electron-transparent sections from Ryugu grain C0071 and three grains of C0369 and analysed them using (scanning) transmission electron microscopy ((S)TEM) (Extended Data Table 1). The chemical compositions of the phyllosilicate matrix in the sections of the four Ryugu grains are plotted along a line with an Mg/(Mg + Fe) atomic ratio of 0.86 and between the serpentine and saponite solid solution lines in the (Si + Al)–Mg–Fe ternary diagram (Supplementary Fig. 3). This is consistent with the compositions of typical phyllosilicate matrices in Ryugu grains reported in previous studies4,13. TEM images show that the sodium-rich veins extend from the grain surface to a depth of ~1 µm and are not mixed with the underlying phyllosilicate matrix (Fig. 2a,b and Extended Data Figs. 2–5). Selected-area electron diffraction (SAED) patterns corresponding to anhydrous sodium carbonate, natrite (Na2CO3), were obtained from the sodium-rich veins in the grain C0071 and the three fine grains in C0369 (Fig. 2c and Extended Data Figs. 2–5). Additionally, diffraction patterns matching hydrous sodium carbonate, thermonatrite (Na2CO3·H2O) were obtained from the vein of one fine grain in C0369 (Fig. 2d). The sodium carbonate veins in all samples consist of polycrystalline structures composed of subgrains (Supplementary Fig. 4). Sections processed by a focused ion beam (FIB) system from sodium-rich patches on the phyllosilicate surfaces in the grain C0071 also consisted of sodium carbonates (Extended Data Fig. 6).

Fig. 2: TEM and STEM analyses of sodium carbonates in Ryugu grains.
figure 2

a, ADF-STEM image of sodium carbonates in Ryugu grain (C0369-113002). The dashed line indicates a boundary between sodium carbonates and the phyllosilicate matrix. b, Composite elemental map of sodium (green), silicon (blue) and iron (magenta) obtained by the STEM-EDX analysis. c,d, SAED patterns of natrite (c) and thermonatrite (d) obtained from the circles in a. e, Elemental compositions of sodium carbonates plotted on a C–O–Na ternary diagram (at.%). Red, ochre, green and blue symbols represent Ryugu samples C0369-113001, C0369-113002, C0369-113003 and C0071, respectively. Filled and open black circles represent stoichiometric compositions of natrite and thermonatrite, respectively.

Full size image

A STEM-energy-dispersive X-ray spectrometry (EDX) analysis showed that the main elements within the sodium carbonates are sodium (Na), carbon (C) and oxygen (O). There are minor amounts of sulfur (S) (<6 at.%) and fluorine (F) (<4 at.%) (Fig. 2, Extended Data Table 1 and Supplementary Figs. 5–8). The average near-edge X-ray absorption fine-structure (NEXAFS) spectrum at the sulfur L-edge in the sodium carbonates of the grain C0071 corresponds to sodium sulfate18 (Extended Data Fig. 6). Assuming that sulfur and fluorine were present as Na2SO4 and NaF, the C/Na and O/Na atomic ratios of the sodium carbonates are ~0.4 to 0.6 and ~1.2 to 1.6, respectively, which are close to the stoichiometric composition of natrite (C/Na = 0.5 and O/Na = 1.5) (Fig. 2e and Methods). The slight variations of their values are challenging to evaluate and interpret because of the limits of the analytical precision of STEM-EDX and possible electron-beam damage during the analysis (Supplementary Fig. 9).

The sodium carbonates in the grain C0071 have many voids with irregular shapes (Fig. 3a). Particulate spots with a high chlorine (Cl) concentration were observed on the surface of the sodium carbonates and the inner walls of the voids (Fig. 3b,c). Each spot is less than 200 nm in size. These particles were confirmed to be crystalline halite (NaCl) by their electron diffraction patterns (Fig. 3d).

Fig. 3: TEM and STEM analyses of halite in Ryugu grain C0071.
figure 3

a, ADF-STEM image of sodium carbonates in Ryugu grain (FIB section C0071-01). The z-contrast is inverted for better visualization. The yellow dashed line indicates a boundary between sodium carbonates and the phyllosilicate matrix. Upper and lower dark areas correspond to a platinum coating (Pt-c). b, Enlarged view of the box region in a. The ADF-STEM image is overlaid with the chlorine distribution (red) obtained by the STEM-EDX analysis. The slight difference in contrast between the right and left sides is due to variations in sample thickness. Relatively large voids are labelled v. c, Bright-field TEM image obtained from the box in b. The circle is the area corresponding to the diffraction pattern in d. d, SAED pattern of a halite particle. BF-TEM, bright-field TEM.

Full size image

Discussion

Formation processes of sodium carbonates

Sodium carbonate minerals are primary precipitates formed by evaporation19 or freeze-out crystallization20 in saline lakes on Earth. Some types of saline lakes are dominated by Na+, HCO3−, CO32− and SO42− and are highly alkaline (pH > 9)19. Chlorine and fluorine are commonly concentrated in alkaline lakes21. As with these environments, sodium carbonates in our Ryugu samples probably formed in alkaline solutions that occurred during Ryugu’s evolution. The aqueous alteration of Ryugu’s parent body began between the initially accreted anhydrous dusts with a CI chondrite composition and CO2-bearing H2O ices when the ices were melted by heating due to the radioactive decay of 26Al within a few million years after the birth of the Solar System3,4,22. Chemical modelling of the aqueous alteration of Ryugu samples and CI chondrites suggests that the initial Mg-Na-Cl solutions evolved toward alkaline brines (pH ≈ 8–10) enriched in Na+ and Cl− and were less abundant in HCO3− and CO32− (refs. 4,23). The main matrix minerals, including phyllosilicates (saponite and serpentine), magnetite, iron- and nickel-bearing sulfides, calcium-magnesium-iron carbonates and calcium phosphate, probably precipitated during the progressive alteration at temperatures less than 100 °C (refs. 3,4,8,24). Our samples are composed mainly of these minerals and, hence, correspond to extensive aqueous alteration. The presence of sodium carbonates alongside the extensively altered lithology suggests that the sodium carbonates precipitated during or after the extensive alteration. The concentrations of HCO3− and CO32− in the chemical model of Ryugu’s aqueous alteration (1.7 × 10−3 mol kg−1 and 1.1 × 10−4 mol kg−1 at 20 °C with a water-to-rock ratio of ~0.9, respectively4) are much lower than the solubilities of the main sodium carbonates, such as nahcolite (NaHCO3) and natron (Na2CO3·10H2O) in the NaHCO3–Na2CO3–H2O solution (~1.2 mol kg−1 and ~2.7 mol kg−1 at 25 °C, respectively) and those in the NaCl–NaHCO3–Na2CO3–H2O solution (~0.16 mol kg−1 and ~2.1 mol kg−1 at 21 °C, respectively)25. This suggests that the amounts of HCO3− and CO32− in the chemical model of Ryugu’s parent body at ≳20 °C are too low for the formation of sodium carbonates. An increase in the concentration of HCO3− and CO32− in the CI chondrite brines could occur through evaporation of water, leading to the precipitation of sodium carbonates23. In addition, sodium carbonates, sulfate and chloride are the main evaporite minerals of saline lakes with a low concentration of calcium and magnesium19. The intensity of water evaporation on the parent body may depend on the ambient pressure. Given that extensive aqueous alteration probably proceeded ~14 km below the surface of the parent body, which had a radius of 65 km (ref. 4), the hydrostatic pressure at that depth has been estimated to be approximately ~0.7 MPa using a grain density of 1.8 g cm−3 (ref. 4, Methods and Supplementary Fig. 10). At this pressure, intense evaporation, such as water boiling, may not have occurred at 100 °C. Instead, if fractures and pores within the parent body were connected to the surface, the venting of H2O would lead to depressurization and evaporation along the fractures26. This water depletion could have happened at a late stage after the main aqueous minerals had precipitated. Evidence of global-scale flows of fluids has been found in the form of carbonate veins on the carbonaceous asteroid Bennu27, whereas such structures remain unclear on Ryugu. Other factors that can control the permeability of fluids include interconnections among micropores and fractures and large-scale convection28.

The cooling of the brines is another possible mechanism that could have caused the precipitation of sodium salts25,29,30. The NaHCO3–Na2CO3–H2O solution has the quadruple point of natron, nahcolite, H2O ice and the solution at approximately −3 °C (ref. 29), whereas hydrated sodium sulfates precipitate at ~0 °C in a dilute Na2SO4–H2O solution30. In the NaCl–NaHCO3 and NaCl–Na2CO3 solutions, natron, nahcolite, halite, hydrohalite (NaCl·2H2O) and H2O ice precipitate at ~0°C to approximately −22 °C (ref. 25). Additionally, thermodynamic calculations have predicted that the cooling of the Na- and Cl-rich brines equilibrated with CI-type rocks would result in the formation of natron at approximately −13 °C, followed by the precipitation of hydrohalite at approximately −22 °C (ref. 23). After Ryugu’s parent body had reached its peak temperature, it cooled due to the exhaustion of the radioactive heat. The remaining alkaline brines probably concentrated as H2O ices formed. As a result, the sodium salts would have formed at subzero Celsius temperatures.

The formation of hydrous sodium salts by the cooling of brines is apparently inconsistent with the presence of anhydrous natrite as the main phase in our samples. The materials on Ryugu may have undergone dehydration processes after the asteroid Ryugu was formed3. These may have included impact heating, solar heating, space weathering and long-term exposure of the asteroid surface3. Previous TEM studies revealed that dehydration and the escape of light elements are notable changes in the aqueous minerals in space-weathered Ryugu samples16,17. Furthermore, experimental and theoretical studies show that hydrous sodium carbonates readily dehydrate and ultimately convert to natrite in vacuum environments31. Considering the space-weathering features in our samples (Extended Data Figs. 1–4), which indicate exposure to the vacuum, it is probable that the sodium carbonates in Ryugu samples were originally hydrated and later experienced some combination of dehydration processes. Halite in Ryugu samples may also have formed as hydrohalite and later dehydrated because it is stable only below ~0 °C (ref. 32). The presence of halite on the surface and pore walls of sodium carbonates (Fig. 3) indicates that the carbonates formed first and then halite formed from liquid flowing through the surfaces and pores of the carbonates. This precipitation order is in line with the cooling of the Na- and Cl-rich brines coexisting with CI-type rocks23,33. If sodium salts precipitated during local or temporal non-equilibrium processes, as proposed for spherulitic magnetite34, the initial hydrate phases and precipitation temperature could differ from those in the equilibrium conditions.

To evaluate the chemical features of the brines, we compared the main elements detected in the sodium carbonates in Ryugu samples with the results from a chemical model of the Na- and Cl-rich brine equilibrated with the main aqueous minerals under the most water-rich condition (water-to-rock ratio of 0.9) in Ryugu’s parent body4, which represents the chemical composition of dissolved ions during extensive aqueous alteration (Fig. 4). The observed sodium carbonates have a chlorine to sodium ratio Cl/Na ≈ 2 × 10−3 and a potassium to sodium ratio K/Na ≲ 10−4, which was under the detection limit. Both values are much lower than those in the model (Cl/Na ≈ 1 and K/Na ≈ 4 × 10−2) (Fig. 4). A plausible process to cause the low concentration of Cl and K is fractional crystallization during cooling of the brines. Liquid at subzero Celsius temperatures can migrate through rock in the direction of decreasing temperature, driven by capillary action under microgravity conditions35. In the flows, most of halite could have formed later at a different location from the sodium carbonates. Other possibilities for the Cl and K depletion include incorporation of K+ into phyllosilicates36. Fluorine is scarce in the chemical model but was detected in the sodium carbonates (Fig. 4). Because not all sodium carbonates contain a detectable amount of fluorine, the fluorine could have been locally concentrated by dissolution from fluorine-bearing minerals. Fluorine could have been removed from apatite (Ca5(PO4)3(OH,F,Cl)) (ref. 3) by the ion exchange mechanism with hydroxide ions37. The presence of sodium sulfate suggests that the alkaline brines are rich in sulfate ions (SO42−). The enrichment of sulfur species in the brines was not predicted in the chemical model, as shown in Fig. 4, but was indicated by an analysis of hot water from Ryugu samples14.

Fig. 4: Elemental compositions of sodium carbonates of Ryugu samples and the chemical model of the Na–Cl brine in Ryugu’s parent body.
figure 4

Amounts of the main elements (F, fluorine, S, sulfur, Cl, chlorine and K, potassium) relative to sodium (Na) contents (at.%). Filled symbols indicate the compositions of sodium carbonates detected by STEM-EDX analysis. Open green squares denote compositions from the chemical equilibrium model of aqueous alteration on Ryugu’s parent body at 40 °C with a water-to-rock ratio of ~0.9 (mass ratio)4.

Full size image

The overall processes forming the sodium salts are summarized in Fig. 5. There were brines enriched in Na+, HCO3−, CO32−, Cl− and SO42− during the last stage of aqueous alteration in Ryugu’s parent body (step 1). If global interconnections of fractures developed, the venting of H2O vapour would cause the precipitation of sodium salts (step 2). Alternatively, when the brines cooled, sodium carbonates and a small amount of sodium sulfate precipitated, followed by the formation of halite (step 3). The sodium-rich surface on Ryugu grain C0071 may have formed due to fracturing along the path of the brines during regolith formation (step 4). The fracturing of the rock may have been facilitated by the positive change in volume of the frozen ice and subsequent compaction of the surrounding phyllosilicates4. Most of the salts covering the surface could have been lost in the harsh space environment, although some could have remained as veins and patches on the grain surface. Hydrated sodium salts may have dehydrated during regolith evolution.

Fig. 5: Evolution of alkaline brines through the history of Ryugu samples.
figure 5

Step 1, Alkaline brines rich in CO32−, HCO3−, Na+, Cl− and SO42− flowed during the final stage of aqueous alteration in the parent body of Ryugu. Large- or small-scale fractures developed in phyllosilicate rocks. Fluorine (F−) was dissolved from apatite. Step 2, If global-scale interconnections of fractures occurred, the venting of H2O vapour towards the surface would promote the precipitation of sodium salts. Step 3, When the brines cooled and H2O ices formed, the remaining brines concentrated, leading to the precipitation of sodium salts. Step 4, During regolith evolution after the formation of Ryugu, sodium-rich surfaces (blue), including sodium carbonate patches and veins, formed by fracturing along the pathway of the brines. In addition, hydrated sodium carbonates were dehydrated. Na-Cb, sodium carbonates; NaCl, halite.

Full size image

Sodium salts found in chondrite meteorites and icy bodies

This study provides evidence of salt precipitation and water activity within Ryugu’s parent body, which occurred at a later stage than previously recognized. Thus far, the Zag and Monahans H chondrites contain halite crystals as exogenous inclusions that probably originated from aqueously altered carbonaceous bodies38,39. Our Ryugu samples present proof that halite exists in carbonaceous chondrite material that has undergone aqueous reactions. Halite has also been reported in the Murchison CM carbonaceous chondrite40 and the Winchcombe CM carbonaceous chondrite41. In these meteorites, halite occurs as euhedral cubes, unlike the slightly elongated halite in the Ryugu sample (Fig. 3c), and has been interpreted as resulting from terrestrial alteration41. Sodium sulfate has been reported in veins in CI chondrite meteorites42, but its origin has been poorly constrained. Our analysis suggests that at least a small amount of sodium sulfate is indigenous to CI chondrite materials. Sodium carbonates have not been identified in CI chondrites, although they share mineralogical and chemical characteristics with Ryugu samples3,4. All CI chondrites ever recovered were probably modified during their long residence on Earth (for example, refs. 3,4,43). Sodium carbonates were probably present in CI chondrites before they fell to Earth but were lost due to terrestrial modifications. No sodium carbonates have been found in other carbonaceous chondrites. Possible reasons for their absence could be as follows: (1) Salt minerals may have been lost through modification on Earth or during sample preparation. (2) Water-to-rock ratios for the main carbonaceous chondrite groups—CM (0.3–0.4), CR (0.1–0.4), CV (0.1–0.2) and CO (0.01–0.1)44—are lower than the ratio for Ryugu (0.2–0.9)4. Their parent bodies probably had less water than Ryugu, resulting in aqueous alteration at higher temperatures and an absence of the long-term fluids necessary for sodium carbonate precipitation. (3) The amount of CO2 in the primordial ice of Ryugu’s parent body was probably higher than that of other carbonaceous bodies. In this case, the concentration of bicarbonate and carbonate ions in brines increased23, making it easier for sodium carbonates to precipitate from the brines. Ryugu’s parent body may have formed beyond the CO2 snow line, possibly closer to cometary formation regions, based on CO2-bearing fluid inclusions4, oxygen and carbon isotopic similarities to comets9,11, and distinct Fe–Ti isotopic compositions compared to other carbonaceous chondrites10. Therefore, Ryugu’s parent body plausibly incorporated more CO2 ices in its formation region.

Sodium carbonates and hydrohalite are expected in surface deposits on the dwarf planet Ceres45,46,47, in water plumes from Saturn’s satellite Enceladus48,49 and on the surfaces of Jupiter’s satellites Europa50 and Ganymede51. The presence of the sodium salts suggests that these icy bodies probably have alkaline subsurface oceans, which formed by aqueous alteration between initially accreted chondrite rocks with CO2-rich ices45,46,47,48,49,50,51. Possible precipitation processes of the sodium salts on Ceres include freezing or evaporation of the alkaline fluids near the surface45,46,47. Thus, the salt production is closely linked to geological settings and brine chemistry in the aqueous bodies. The discovery of sodium salts in Ryugu samples may provide new insights for comparing the evolution of water in carbonaceous bodies and alkaline subsurface oceans in the icy bodies.

Methods

Sample preparation and analytical procedure

Ryugu samples from the second touchdown sites were preserved in chamber C of the sample catcher inside the sample container of the Hayabusa2 spacecraft. One coarse grain, C0071 (1.54 mm maximum diameter), and aggregates of fine grains, C0369 (grain size less than 1 mm across), were used in this study. These grains were preserved in the clean chambers used for Ryugu samples at the Extraterrestrial Sample Curation Center of JAXA and were allocated to the authors as part of the international announcement of opportunity for Hayabusa samples. A reflected light image of the grain C0071 (Fig. 1a) was obtained using an optical microscope (JASCO IRT-5000) in a nitrogen-filled glovebox at JAXA. The grain C0071 was analysed by field-emission scanning electron microscopy (FE-SEM) at JAXA and by SR-XRD and SR-computed tomography (CT) at SPring-8 without being exposed to air. These samples were subsequently observed using FE-SEM and (S)TEM at Kyoto University and by SR-based scanning transmission X-ray microscopy (STXM) at UVSOR. Samples were stored in a dry glovebox filled with nitrogen at Kyoto University. Nitrogen-filled vessels were used to transport samples. The grain C0071 was exposed to the atmospheric environment during sample exchange processes for the FE-SEM observations, FIB processing and (S)TEM analysis at Kyoto University. The total exposure time was less than 2 min. The assembly of grains C0369 was handled in a nitrogen-filled glovebox at Kyoto University. Using the micromanipulation systems equipped in the glovebox, approximately 200 grains were fixed onto carbon-conductive sheets. Then, FE-SEM observations, FIB processing and (S)TEM analysis were conducted for these grains. During sample exchange processes for these analyses, the grains were exposed to air for less than 2 min.

SR-XRD and SR-XCT analyses

The SR-CT and SR-XRD analyses were performed at beamline BL20XU of SPring-8, Japan. The grain C0071 was enclosed in a capsule made of polyimide film, which allows X-rays to pass through the sample while preventing exposure to air during imaging. The X-ray energy used in this study was 37.7 keV. The three-dimensional internal structure of the grain was investigated with the CT mode. The X-ray detector for the CT mode was equipped with a scintillator consisting of gadolinium aluminium gallium garnet (Gd3Al2Ga3O12:Ce), which was 20 µm thick, a complementary metal oxide semiconductor (CMOS) camera with 2,048 × 2,048 pixels (Orca Flash 4.0) and a ×10 lens. The pixel size of the CT mode was 0.494 µm and the field of view was ~1.78 mm with the offset CT scan mode. The typical mineral phases in the grain were determined with the XRD mode. An X-ray beam probe focused with a Fresnel zone plate was used in the XRD mode. The X-ray detector for the XRD mode was equipped with a scintillator consisting of a P43 (Gd2O2S:Tb) that was 100 µm thick, a CMOS camera with 2,048 × 2,048 pixels (Orca Flash 4.0) and relay lenses. The detector was placed 121 mm behind the sample. Diffraction images of 2θ ranged from 0.92° to 16.33° (lattice spacing d = 20.5–1.16 Å). The pixel size of the SR-XRD system was 19.23 µm.

SEM analysis

Surface features of the grain C0071 were observed by FE-SEM (Hitachi SU6600) at JAXA. The sample was transported to the FE-SEM using a sample transfer vessel immediately after the allocation and was not exposed to the atmosphere before the SEM observations. Secondary electron images were obtained at an acceleration voltage of 2 kV with an electron-beam current of less than 100 pA. Elemental mapping analyses were performed by EDX using the FE-SEM equipped with an X-MaxN 150 mm2 (Oxford Instruments). The electron beam, which had an acceleration voltage of 15 kV and an electron-beam current of ~100 pA, was applied for EDX analysis and backscattering electron imaging. Surface features of ~200 grains from C0369 and C0071 were observed using FE-SEM (JEOL JSM-7001F) at Kyoto University, before FIB processing. A nitrogen-filled container was used to transfer a sample from the glovebox to the SEM. Samples were exposed to air for less than 10 s during the sample exchange procedure for SEM. Secondary electron images were obtained at an acceleration voltage of 2 kV and an electron-beam current of ~15–80 pA. Elemental mapping analyses were performed using an X-MaxN 150 mm2 (Oxford Instruments). An electron beam with an acceleration voltage of 15 kV and an electron-beam current of ~100 pA was applied in the EDX analysis and backscattering electron imaging.

Sample preparation using a FIB system

Electron-transparent sections of regions of interest on the Ryugu grains were extracted for (S)TEM and STXM studies using a FIB system (Helios NanoLab G3 CX) at Kyoto University. A container filled with nitrogen or under vacuum was used to transfer the sample to the FIB system. The grain C0071 was fixed with indium for the FIB processing. Samples were exposed to air for less than 10 s during the sample exchange procedure for the FIB system. Surface features were observed before the FIB processing using an acceleration voltage of 2 kV and an electron-beam current of ~40–80 pA. To protect the grain surfaces during FIB processing and enhance the electrical conductivity, Ryugu grains were coated with a Pt layer deposited by an electron beam (at 2 kV) and by a Pt layer deposited by a Ga-ion beam (at 30 kV). For each Ryugu grain, sections that were a few tens of micrometres in size were extracted using a microsampling manipulator and mounted onto TEM copper grids. After the target samples were attached to the TEM grids, they were thinned to 50 to 200 nm using a 16–30 kV Ga+ beam and were finally cleaned using a 2 kV Ga+ beam at 77 pA.

TEM analysis

The prepared sections were examined using an FE-TEM (JEOL JEM 2100 F) equipped with an EDX spectrometer (JEOL, JED-2300T) at Kyoto University. Samples were exposed to air for less than 1 min while being fixed to the TEM holder. Bright-field, dark-field and high-resolution TEM images as well as SAED patterns at 200 kV were obtained using a CCD or CMOS camera (Gatan Orius200D, Rio9). The mineral phases were identified from the electron diffraction patterns using the software ReciPro52 (Supplementary Table 1). The EDX analysis was performed in STEM mode aided by annular dark-field (ADF) imaging. To analyse the spatial variation of the diffraction patterns, a scanning precession electron diffraction analysis with beam precession was performed using the NanoMegas ASTER/Topspin acquisition system. The virtual dark-field images obtained by scanning precession electron diffraction were analysed using the software Gatan DigitalMicrograph and the software ImageJ. Quantitative elemental abundances were calculated using the ζ-factor method53. EDX data were analysed using the JEOL analytical station software. In addition to the analysis of Ryugu samples, reagent-grade anhydrous sodium carbonate (Na2CO3) (Hayashi Pure Chemical Ind., Ltd) was analysed by STEM-EDX. The ternary diagram of sodium carbonates in Ryugu samples shown in Fig. 2e was calculated from the chemical compositions in Extended Data Table 1. To estimate the C, O and Na compositions of sodium carbonates, we assumed that sulfur, fluorine and chlorine were present as Na2SO4, NaF and NaCl in the sodium carbonates, respectively. We then estimated the amount of Na and O included in Na2SO4, NaF and NaCl and subtracted these values from the total amount of Na and O. The ternary diagram was made using the software Origin.

STXM analysis

NEXAFS spectra of the carbon K-edge and sulfur L-edge of FIB sections of C0071 were measured using the STXM beamline54, BL4U, at the UVSOR Synchrotron Facility, Institute for Molecular Science (Okazaki, Japan). The custom-built sample transfer vessel and a nitrogen-filled glovebox were used to prevent a sample from being exposed to air. After mounting the sample in the STXM system, the chamber was evacuated (down to 0.3 mbar) and backfilled with helium gas at 20 mbar. The incoming X-ray beam was focused onto the sample by a Fresnel zone plate of size 50 × 50 nm2. The sample was scanned in two dimensions, and the transmitted X-ray intensities in each pixel were detected by a photomultiplier tube with a scintillator (P-43). Then, by changing the energy, a two-dimensional spectrum (namely, the ‘energy stack’) was obtained. The energy step sizes for the carbon K-edge NEXAFS spectra were 0.25 eV in the pre-edge region (280–284 eV), 0.1 eV in the first near-edge region (284.1–290 eV), 0.2 eV in the second near-edge region (290.2–295 eV) and 0.4 eV in the post-edge region (295.4–299.8 eV). The energy step sizes for the sulfur L-edge NEXAFS spectra were 0.4 eV in the pre-edge region (160–163.2 eV), 0.2 eV in the first near-edge region (163.4–175 eV), 0.3 eV in the second near-edge region (175.2–185.1 eV) and 0.5 eV in the post-edge region (185.5–190 eV). The acquisition time per image pixel for each energy step was set to 2 ms. The data were analysed using the software aXis 2000.

Estimating the hydrostatic pressure in the parent body of Ryugu

The parent body was assumed to be spherically symmetric. For simplicity, the density was assumed to be constant regardless of the radius. The pressure under hydrostatic equilibrium can be expressed as

$$\frac{\mathrm{d}P(r)}{\mathrm{d}r}=-\rho g(r),$$

where r denotes the radial distance from the centre of the body, P represents the pressure at radius r, ρ is the density and g(r) is the gravitational acceleration at radius r. Given that the pressure is zero at the surface of the body, P can be described as follows:

$$P(r)=\frac{2\uppi }{3}G{\rho }^{2}({R}^{2}-{r}^{2}),$$

where R is the radius of the parent body and G is the gravitational constant. When we assumed ρ = 1.8 g cm−3 (ref. 4) and R = 65 km, then P ≈ 7.4 × 105 Pa at r = 51 km (14 km below the surface).