Oxygen isotope evidence from Ryugu samples for early water delivery to Earth by CI chondrites

The delivery of water to the inner Solar System, including Earth, is still a debated topic. A preferential role for hydrated asteroids in this process is supported by isotopic measurements. Carbonaceous chondrite (CC) meteorites represent our main source of information about these volatile-rich asteroids. However, the destruction of weaker materials during atmospheric entry creates a bias in our CC data. The return of surface materials from the C-type asteroid 162173 Ryugu by the Hayabusa2 spacecraft provides a unique opportunity to study high-porosity, low-density, primitive materials, unrepresented in the meteorite record. We measured the bulk oxygen isotope composition from four Ryugu particles and show that they most closely resemble the rare CI (CC Ivuna-type) chondrites, but with some differences that we attribute to the terrestrial contamination of the CI meteorites. We suggest that CI-related material is widespread among carbonaceous asteroids and a more important source of Earth’s water and other volatiles than its limited presence in our meteoritic collection indicates. Oxygen isotopic measurements from the Ryugu returned sample suggest that primitive hydrated carbonaceous chondrites are an important source of early water and other volatiles to Earth, despite being underrepresented in our meteorite collection due to their preferential destruction during atmospheric entry.

The delivery of water to the inner Solar System, including Earth, is still a debated topic. A preferential role for hydrated asteroids in this process is supported by isotopic measurements. Carbonaceous chondrite (CC) meteorites represent our main source of information about these volatile-rich asteroids. However, the destruction of weaker materials during atmospheric entry creates a bias in our CC data. The return of surface materials from the C-type asteroid 162173 Ryugu by the Hayabusa2 spacecraft provides a unique opportunity to study high-porosity, low-density, primitive materials, unrepresented in the meteorite record. We measured the bulk oxygen isotope composition from four Ryugu particles and show that they most closely resemble the rare CI (CC Ivuna-type) chondrites, but with some differences that we attribute to the terrestrial contamination of the CI meteorites. We suggest that CI-related material is widespread among carbonaceous asteroids and a more important source of Earth's water and other volatiles than its limited presence in our meteoritic collection indicates.
Between June 2018 and November 2019, the JAXA Hayabusa2 spacecraft made detailed spectroscopic observations and measurements of the C-type asteroid 162173 Ryugu. Material from two different locations on the asteroid was collected and returned to Earth on 6 December 2020 (ref. 1 ). One sample was stored in Chamber A of the return capsule and the other, collected close to an impactor-formed crater, was stored in Chamber C. Near-IR spectroscopic data obtained during orbital observations of asteroid Ryugu indicated that it was composed of material "similar to thermally and/or shock-metamorphosed carbonaceous chondrite meteorites" (refs. 2,3 ), with a potential match to the CY (Yamato-type) chondrites 4 . In contrast to this interpretation, initial curation studies at the JAXA ISAS facility suggested that the returned Article https://doi.org/10.1038/s41550-022-01824-7 by secondary ion mass spectrometry has revealed a large variation in δ 18 [9][10][11]. As the dominant phase in Ryugu particles (64 to 88 vol.%), phyllosilicates are likely to have a δ 18 O composition that is relatively close to the mean bulk value of 15.88‰ determined in this study (Table 1). This value is within the range determined for CI matrix separates 12 . In view of the heterogeneity of Ryugu particles and the wide variation in δ 18 O displayed by different mineral phases, the range of values measured in this study is not unexpected. Where bulk oxygen isotope values have been determined on mg-sized fractions taken directly from CC meteorites, it is common to obtain a range of δ 18 O values similar to, or exceeding, those obtained in this study (Methods).
Calculations based on measured modal data for Ryugu particles 7 and oxygen isotope analysis of Ryugu mineral phases 10 yielded bulk δ 18 O values between 9.7 and 18.9‰ (Supplementary Information), which is close to the range determined in this study. The overall range of δ 18 O values measured in Orgueil (14.39 to 16.62‰) and Y-82162 (20.77 to 24.47‰) is greater than would be anticipated on the basis of our measured system precision (±0.1‰). These meteorites were run as relatively coarse-grained powders to reflect the overall grain size of the Ryugu material and were not ground to very fine powders as required for complete homogenization. Intense grinding would probably modify their primary compositions, for example there is the potential for substantial changes to the water content of these hydrated samples.
The single particle analysed from chamber A (A0098) has the lowest δ 18 O value (11.46 ± 0.12‰ (2 s.d.)) of all the Ryugu material analysed in this study, and one of the lowest Δ 17 O values (0.56 ± 0.06‰ (2 s.d.)). However, it is also one of the smallest samples analysed here, and as such one of the most susceptible to the effects of sampling a heterogeneous mineralogy. Other small samples also show some of the largest variations, for example C0014-3 and C0087 with Δ 17 O values of 0.54 and 0.75, respectively (Table 1). Therefore, the most likely explanation for the low δ 18 O value of A0098 is sample heterogeneity, possibly reflecting a higher modal magnetite content. samples were "most similar to CI chondrites" (ref. 1 ). These contradictory classifications can only be resolved by detailed characterization studies of the Ryugu particles. In particular, high-precision oxygen isotope analysis is widely recognized as the most powerful technique for establishing the interrelationships between individual samples and well-characterized meteorite groups. The results presented here provide a firm basis for evaluating the relationship between the Ryugu samples and the carbonaceous chondrite (CC) meteorite inventory.

Results
Subsamples from four distinct Ryugu particles were analysed for their bulk oxygen isotopic compositions by laser fluorination, using a 'single-shot' technique 5,6 (Methods). Three of the four analysed samples were from Chamber C (C0014,21; C0068,21; C0087,2) and one from Chamber A (A0098,2). Sample transport, loading and analysis techniques used in this study ensured that at no time were the particles exposed to atmospheric contamination (Methods).
The four particles from which the analysed material was extracted consist predominantly of fine and coarse-grained phyllosilicates, varying between approximately 64 and 88 vol% (ref. 7 ) (Figs. 1a,b). Anhydrous silicates (olivine and pyroxene) have not been observed in any of these four particles, but rare examples have been identified in other Ryugu particles 8,9 . Phyllosilicates comprise a serpentine-saponite intergrowth and have bulk compositions that fully overlap with those found in CIs 7 . Carbonate minerals, mainly dolomite, with minor Ca-carbonate and breunnerite, are present in highly variable amounts (approximately 2 to 21 vol%) 7 . Magnetite (approximately 3.6 to 6.8 vol%), as framboids, plaquettes and spherical aggregates, and sulfide minerals (approximately 2.4 to 5.6 vol%) are also present within the phyllosilicate-rich matrix (Fig. 1a,b) 7,9 . Ryugu particles have a high average porosity of 41% and consequently a low average density of 1,528 ± 242 kg m −3 , comparable to that of the CI chondrite Orgueil or the ungrouped primitive meteorite Tagish Lake 9 .
Seven individual analyses were undertaken on material extracted from the four Ryugu particles (Methods). The mass of material analysed varied from 0.18 to 1.83 mg (Table 1). For comparison with the Ryugu particles, the CI chondrites Orgueil, Ivuna and Alais and the CY chondrites 4 Y-82162 and B-7904 were also analysed as part of this study (Table 1). Due to the wide range of Ryugu particle masses analysed in this study, a weighted average has been calculated for comparison purposes with the CY and CI chondrites (Table 1). Unless otherwise specified, in the text and figures, the weighted Ryugu composition has been compared to the unweighted averages for the CIs and CYs. Weighted and unweighted average data for all samples analysed in this study are given in Table 1.
The mean oxygen isotopic composition (weighted) of the seven Ryugu analyses are shown in Fig. 2 along with data for potentially related CC groups. The Ryugu particles have a mean oxygen isotope composition that overlaps with that of the CI chondrites, but is considerably lighter with respect to δ 18 O than the CYs (Fig. 2). A possible match between Ryugu samples and CIs has also been suggested on the basis of bulk oxygen isotope data in two other recent studies 9,10 .

δ 18 O variation in Ryugu particles
Individual Ryugu analyses show a large range in δ 18 O values, from 11.46 to 19.30‰ (Table 1 and Fig. 3). The largest subsample assigned for oxygen isotope analysis came from particle C0014 and had a total initial mass of 5.5 mg, permitting multiple measurements (n = 4) (total analysed mass so far 3.3 mg) (  7,9 show a considerable level of heterogeneity within individual Ryugu particles (Fig. 1a,b). Analysis of individual mineral phases in Ryugu particles Article https://doi.org/10.1038/s41550-022-01824-7

Ryugu analyses compared to CI and CY chondrites
Despite the large variation in δ 18 O values displayed by the Ryugu particles, they tend to cluster close to our analyses of Orgueil and Alais in Fig. 3. Ivuna has a δ 18 O composition that falls within the Ryugu range, but has a lower Δ 17 O value. The Ryugu particles are isotopically distinct from the CY chondrites in respect of both δ 18 O and Δ 17 O (Fig. 3). The two CYs analysed in this study (Y-82162 and B-7904) have similarly high δ 18 O values, but their Δ 17 O values are distinct. This suggests that the CYs, as currently defined 4 , are not a single homogeneous group. This finding merits further investigation but does not alter the principal conclusion of this study that Ryugu particles are closely related to CI chondrites. This similarity is well demonstrated when the weighted mean composition of the Ryugu particles (weighting by mass of O 2 gas liberated during fluorination) is compared to the mean CI value (Fig. 3 is lower than the weighted mean value for the Ryugu particles (0.66 ± 0.09‰ (2 s.d. weighted)), but there is clearly significant overlap at the 2 s.d. level (Fig. 3).
Ryugu particles analysed in other studies 9,10 overlap, or have similar oxygen isotope compositions to those obtained here, with the notable exception of one analysis 10 that lies at the edge of the CY* field in Fig. 4. This value is not thought to reflect analytical differences with the other laboratory involved in this joint study, but instead is attributed to small scale intrinsic heterogeneity within the Ryugu regolith 10 . Previous oxygen isotope analyses of CY chondrites 13 provide additional support for the possibility that these meteorites respresent two distinct groups (Fig. 4).

Discussion
It is clear from our oxygen isotope data that there is a much stronger case for a connection between the Ryugu particles and CIs than with CYs (Figs. [2][3][4]. This potential relationship is also supported by detailed mineralogical and petrological studies of Ryugu material 7,9,10 . Differences between the Ryugu particles and CIs that have been identified so far probably reflect terrestrial alteration of the latter. It is well documented that Orgueil, which fell in 1864, has undergone considerable mineralogical modifications due to terrestrial weathering 14 and this would necessarily result in the incorporation of atmospheric oxygen and so pull the bulk Δ 17 O value closer to the terrestrial fractionation line (TFL). This conclusion is in keeping with the mineralogical evidence that Ryugu particles do not contain ferrihydrite or sulfate 7,10 , whereas Orgueil does 15 . There is also evidence that phyllosilicates in at least some Ryugu samples may lack interlayer water in the saponite component 10 . Stepwise pyrolysis of Orgueil has demonstrated that the interlayer water present in the meteorite is of terrestrial origin, as is also the case for the more recent CC fall Tagish Lake 16 .
All the CI chondrites measured in this study have lower mean Δ 17 O compositions (Alais 0.60 ± 0.01‰; Ivuna 0.41 ± 0.01‰; Orgueil 0.58 ± 0.08‰) than the weighted mean Ryugu value (0.66 ± 0.09‰ 2 s.d.) (Table 1). We have undertaken calculations using our analyses of Orgueil to examine the possibility that the Δ 17 O difference between the Ryugu particles and CIs is the result of terrestrial contamination of the latter (Supplementary Information). Using either the modal composition of Orgueil 15 (method 1), or its full chemical analysis 17 (method 2), these calculations indicate that the measured Δ 17 O value of Orgueil (0.58‰) can be fully accounted for in terms of the terrestrial contamination of material with a pre-atmospheric composition identical to that of the Ryugu particles (0.66‰). It is important to note that these calculations do not provide unequivocal confirmation that the Δ 17 O differences between the CIs and Ryugu is solely the result of terrestrial contamination of CI chondrites and it remains possible that primary differences between these materials may also be a factor. However, the evidence that Ryugu grains lack interlayer water 10 and that such water in CIs may be of terrestrial origin 16 is consistent with this Δ 17 O difference being the result of terrestrial contamination.
The larger Δ 17 O difference between Ivuna and Ryugu particles compared to the other CIs may reflect local-scale heterogeneity. Studies have shown that CIs display chemical heterogeneity at sampling scales of less than 1 to 2 g (refs. 18,19 ). Such large samples of these important meteorites are rarely available and homogeneous powders are normally based on 100 to 200 mg aliquots. However, despite the possibility of local-scale heterogeneities, all three CI meteorites measured here display lower Δ 17 O values than the weighted average for Ryugu. This is consistent with the other lines of evidence indicating that they have experienced a substantial degree of terrestrial contamination. This may have important implications for the use of CI meteorite bulk compositional data as proxies for Solar System values 20 . In addition, as a result of adding a large terrestrial water component, and the realistic possibility of contamination by terrestrially-derived organic molecules, light stable isotope data (C, H, O, N) from CI meteorites needs to be carefully evaluated, as it is likely to include an important non-indigenous component. As a result of their lack of terrestrial contamination, chemical and isotopic data from Ryugu samples will provide a fresh perspective on these bulk Solar System values.
While only 5.4 g of material was collected by the Hayabusa2 spacecraft, initial spectral characterization of the returned samples indicated that they provide a good match to the global average data obtained during orbital observations of Ryugu 1 . Hence, the returned particles are probably representative of the asteroid as a whole. The close match between δ 18 O compositions of Ryugu particles and the CIs and the likelihood that both had very similar preterrestrial Δ 17 O values provides a firm basis for linking asteroid Ryugu to the CI chondrites.
CI chondrites are a rare group of meteorites, with only nine examples (November 2022) listed on the Meteoritical Bulletin database 21 , of which four are probably members of the CY group 4 . This compares to 724 entries (November 2022) for CM2 (Mighei-like) chondrites, the most abundant group of hydrated CCs. However, the apparent paucity of CI-related material arriving on Earth may simply reflect their low-strength characteristics 22 . While CC meteorites represent only about 4% of observed meteorite falls (Meteoritical Bulletin Database) 21 they comprise 55-60% of the micrometeorite population (fragments in the size range 10 μm-2 mm sized), which accounts for most of the 40,000 ± 20,000 metric tons of extraterrestrial material accreted by the Earth each year 23 . CI-related particles have been tentatively identified within the larger-sized fraction of micrometeorites and may be more common among the less well studied, smaller-sized particles 23 .
CI chondrites have very short cosmic ray exposure ages, which are generally less than 2 Myr (ref. 24 ). Asteroid Ryugu is probably the product of multiple parent body disruption/resurfacing events, but is estimated to have formed in its present 'spinning-top' shape more than 8.5 Myr ago 25 . This raises the possibility that Ryugu may represent the immediate source body to the CIs, including the important meteorites Orgueil, Ivuna and Alais. Ryugu is an Apollo Earth-crossing asteroid with an aphelion of 1.419 astronomical units (AU) and a perihelion of 0.96 AU (ref. 26 ). In contrast, calculations of the pre-atmospheric orbit of the Orgueil meteorite suggest an aphelion beyond the orbit of Jupiter 27 . Pre-atmospheric trajectories determined for recent CC falls all have aphelions in the outer main belt 28 , consistent with the proposal that near-Earth objects are not a major source of meteorites, with most falls originating directly from the main belt 29 . It therefore seems unlikely that the known CI meteorites originated from Ryugu.
Ryugu is classified as a Cb-type asteroid, possibly derived from either the Eulalia or Polana asteroid families 30 . Bennu, the target asteroid of the OSIRIS-REx mission, is also probably derived from one of these two asteroid families 31 . CI chondrites have been spectrally matched to C, Cb and B-type asteroids 32 , which make up about half of all the C-complex bodies in the main belt 30 . These types are also well represented in the inner main belt and probably supply a notable fraction Article https://doi.org/10.1038/s41550-022-01824-7  7 and ** from Greenwood et al. 6 . $ mass was estimated based on measured 17% yield for particle C68 -see methods for further details.   17 . Even after subtraction of the potentially contaminated interlayer water, CI chondrites have water contents higher than CM2s (ref. 17 ). While CM chondrites show a strong link to Ch asteroids, they also have affinities to all other C-complex classes 32 . A mix of CI and CM chondrites seems a strong possibility for hydration of the inner Solar System, with Ryugu data pointing to a more important role for CIs than their paucity as meteorites might suggest. How the Earth gained its water remains an outstanding issue in planetary science 33 . While a small fraction may have been inherited from the protosolar nebula, modelling and isotopic studies suggest this may only be about 1%, with the remaining 99% delivered by CCs during the main phase of Earth's accretion [33][34][35] . CIs, as well as CMs, have H and N isotopic compositions close to that of the bulk Earth, whereas cometary sources are isotopically far removed from terrestrial values 35 . CIs are the only meteorite group that shows a close match to Earth's nucleosynthetic Fe isotope composition (μ 54 Fe) 36 . On the basis of the evidence from Ryugu discussed above, the uncontaminated Δ 17 O composition of CIs of 0.66‰ makes them the hydrated CC group with the closest oxygen isotopic composition to Earth. However, there remains considerable uncertainty about how much water is present on Earth and it is possible that other meteorite groups contributed to its water budget. Enstatite chondrites have been proposed as a potential source of Earth's water 37 . However, indigenous water in enstatite chondrites is substantially lower than in CCs 17 . So, while enstatite chondrites could have supplied up to three ocean masses (one ocean mass 1.38 × 10 21 kg), this would be at the lower end of terrestrial water estimates, which might be up to 18 ocean masses 38 , with experimental evidence indicating considerable amounts of hydrogen partitioned into Earth's core 39 .
CI chondrites, and CI-related asteroids such as Ryugu, are highly altered (Fig. 1), having experienced extensive parent body hydrothermal processing, such that only traces remain of their original silicate mineralogy [7][8][9]15,18,19 . Despite this, CIs are chemically the most primitive CC group, with a bulk composition close to that of the solar photosphere for most elements [18][19][20] . Although local-scale heterogeneities are present in CIs 18,19 , their bulk chemical composition is essentially unfractionated, suggesting that aqueous alteration took place isochemically, potentially under static fluid conditions 40 . In contrast, modelling studies point to major fluid migration and hence open-system behaviour 41 . Recent measurements made on trapped fluid in a Ryugu pyrrhotite crystal indicate that it contains halogens, nitrogen, sulfur, CO 2 and dissolved organic compounds 42 . As in terrestrial systems 43 , high solute concentration may have been important in controlling fluid flow by reducing the density contrast with the enclosing silicate material. In addition, the primordial Ryugu parent body may have been small, potentially no more than about 20 km in diameter 11 . Models of CC alteration are generally based on much larger bodies, for example of 50 km radius 44 . Alteration on a small asteroid in which the fluid was stagnant may resolve the contradictions between studies that favour isochemical alteration 18,19,39 and the open-system behaviour predicted by numerical simulations 40 .
It has been proposed that Ryugu may be of cometary origin 9,45 , as has also been suggested for Orgueil 27 . Direct measurement of the gaseous atmosphere of Comet 67P/Churyumov-Gerasimenko by the Rosetta spacecraft gave a δ 18 O value of close to 120‰ (ref. 46 ). While this measurement is subject to very large errors, it suggests that cometary ice is very 16 Table 1. Other data are CO3 chondrites 57 (blue circles), CM2 (red squares) and C2 ungrouped (black squares) 13,58-61 , along with analyses for Tagish Lake (inverted red triangle) and Sutter's Mill (pink square) (Supplementary Information). The red line is the best fit line through CM2 (finds and falls) data only. CCAM, CCs anhydrous minerals line 13 . Value n refers to the number of individual aliquots of material that were independently run on the laser fluorination line for each sample (Table 1).

Fig. 3 | Oxygen isotope composition of the samples analysed in this study.
Ryugu particles (green squares) show a wide variation in δ 18 O values, which reflects intrinsic isotopic heterogeneity at the sampling scale involved (see text for further discussion). CI chondrites (coloured diamonds) and CY chondrites (coloured triangles) were analysed for comparison purposes with the Ryugu particles. In terms of their δ 18 O values, the weighted average value for the Ryugu particles and the mean value for the CIs (Alais, Ivuna and Orgueil) are very close in composition (Table 1). In contrast, the mean value for the CY chondrite Y-82162 is substantialy displaced to higher δ 18 O values compared to either the Ryugu particles or CIs. CM2 Line is the extension of the best fit line through CM2 falls and finds shown in Fig. 2. Error bars ±2 s.d.

Article
https://doi.org/10.1038/s41550-022-01824-7 (Fig. 5). In contrast, the final fluids on the CI/Ryugu parent body/bodies would have evolved to more 16 O-rich values, due to protracted exchange with 16 O-rich solids 8,9 . There is little evidence that such extensive levels of aqueous alteration have taken place within cometary nuclei, with particles from Comet 81P/Wild 2 sampled during the Stardust mission dominated by a 16 O-rich, high temperature assemblage 49 . The evidence from Ryugu and CIs is that their parental sources were early-formed asteroids 7,9,11 that underwent extensive aqueous alteration, in response to the decay of short-lived radionuclides, such as 26 Al (t 1/2 = 0.73 Myr) 43 .
Orgueil contains 10.8 wt% structurally-bound water 17 which, on the basis of the results of stepwise pyrolysis 16 , is likely to be of extraterrestrial origin. As discussed earlier, the magnitude of Earth's water inventory is poorly constrained 33,37 . Taking a median estimate of ten ocean masses 33 requires a CI contribution of 2.1% to Earth's mass, equivalent to 54× the mass of the asteroid belt. There is debate about when such material was added to Earth 50 . Modelling studies suggest that water could have been added to Earth throughout its formation, with smaller bodies involved in the earlier stages and a few larger, late-accreting bodies, delivering the bulk of the water during the final stages of terrestrial formation 50 . O, Ru and Mo isotopic evidence suggests that CC material was not added any later than the Moon-forming giant impact 51,52 ,. On the basis of Mo isotopic evidence, it has been proposed that much of Earth's water was delivered during the Moon-forming event by an impactor of CC composition 53 .
Water delivery to the inner Solar System, including Earth, appears to have taken place in three, probably overlapping, stages: (1) early nebular ingassing 32,38 , (2) delivery by small asteroidal bodies, possibly in response to giant planet migration 50 and finally, (3) effects of giant protoplanets, possibly of CC composition, during the final stages of Earth's accretion 50,52 . It would have been during stage 2 that CI-related bodies, which were small and probably originated in the outer Solar System 7,54 , would have made their most important contribution to the hydration of the inner Solar System, including Earth.

Methods
Oxygen isotopic analysis was undertaken at the Open University (Milton Keynes, UK) using an infrared laser-assisted fluorination system 6 . Four distinct Ryugu samples were transported to the Open University in two sealed, nitrogen-filled FFTC (facility-to-facility transport containers). One of the two FFTC contained grains from the initial Hayabusa2 touchdown collection (particle A0098,2, five grains), the other FFTC contained three sets of particles from the second, post-impactor collection: C0014,2 one particle 5.5 mg; C0068,2 one particle 0.5 mg and C0087,2 approximately ten grains, 0.8 mg. Both holders were stored at the Open University in a dedicated cabinet with a continuously purged nitrogen atmosphere.
Sample loading was undertaken in a nitrogen 'glove box' with monitored oxygen levels below 0.1%. A new Ni sample holder was fabricated for the Ryugu analysis work that consisted of just two sample wells, one for the Ryugu particles and the other for the internal obsidian standard. During analysis, the sample well containing the Ryugu material was overlain by a 1-mm-thick, 3-mm-diameter internal BaF 2 window to retain the sample during laser reaction. The flow of BrF 5 to the sample was maintained by gas mixing channels scribed into the Ni sample holder. The sample chamber configuration was also modified so that it could be removed from the fluorination line under vacuum and then opened within the nitrogen-filled glove box. The two-part chamber was made vacuum tight using a compression seal with a copper gasket and quick-release KFX clamp 6 . A 3-mm-thick BaF 2 window at the top of the chamber allowed simultaneous viewing and laser heating of samples. Following sample loading, the chamber was reclamped within the filled nitrogen glove box and then reattached to the fluorination line. Before analysis, the sample chamber was heated overnight under vacuum to a temperature of about 95 °C to remove any adsorbed moisture. Following overnight heating, the chamber was allowed to cool to room temperature and then the flexi section that had been brought up to atmosphere during the sample transfer process was purged using three aliquots of BrF 5 to remove any moisture. The oxygen isotope composition of these 'flexi' blanks were analysed using the MAT 253 micro-volume facility. These procedures ensured that the Ryugu samples were never opened to the atmosphere or contaminated with moisture from those parts of the fluorination line that had been brought up to atmosphere during the sample loading procedure.
All Ryugu samples were run in modified single-shot mode 5 . This procedure involved a single 5 min chamber blank to reduce and eliminate any residual moisture adsorbed on to the sample chamber walls. The oxygen isotope composition of this blank was analysed using the MAT 253 micro-volume facility. Following this blank analysis, the sample itself was run. Sample heating in the presence of BrF 5 was carried out using a Photon Machines Inc. 50 W infrared CO 2 laser (10.6 μm) mounted on an X-Y-Z stage. Reaction progress was monitored by means of an integrated video system. After fluorination, the released O 2 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.
For five of the seven Ryugu samples, the amount of O 2 gas liberated during reaction was much 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 micro-volume. For monitoring purposes, a post-reaction blank was then run and its oxygen isotope composition was also determined. Finally, the internal obsidian standard was fluorinated and analysed. The gas liberated during the 5 min 'prereaction' blank procedure invariably had a composition close to the TFL indicating that it was predominantly composed of residual adsorbed atmospheric moisture.
The NF + fragment ion of NF 3 + can cause interference with the mass 33 beam ( 16 O 17 O). To eliminate this potential problem all samples were treated using a cryogenic separation procedure. This was either done in the forwards sense before analysis on the MAT 253, or as a second

Fig. 4 | Plot of δ 18 O versus Δ 17 O for samples analysed in this study (coloured symbols) compared to the results obtained in other studies (grey symbols).
Fields for Ryugu particles (green), CI (blue) and CY (yellow and mauve) chondrites are based only on the analyses obtained in this study (Fig. 3). Ryugu particles analysed in other studies (grey squares) 9,10 overlap, or are close to, those obtained here, with the exception of one analysis that lies at the edge of the CY* field 10 . Also shown are earlier analyses of CI and CY chondrites 13 (grey diamonds and triangles). CY chondrites appear to represent two distinct groups with similar δ 18 O values, but distinct Δ 17 O compositions. As a consequence, the respective CY fields have been labelled CY and CY*. CM2 Line is the extension of the best fit line through CM2 falls and finds shown in Fig. 2. Article https://doi.org/10.1038/s41550-022-01824-7 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 by raising the temperature of −130 °C. Extensive tests have shown that NF + is retained on the first molecular sieve and that little or no fractionation results from the use of this technique.
Overall system precision in bellows mode, as defined by replicate analyses of our internal obsidian standard (n = 38), is: ±0.05‰ for δ 17 O; ±0.10‰ for δ 18 O; ±0.02‰ for Δ 17 O (2 s.d.) 55 . Overall system precision in micro-volume mode is slightly lower than in bellows mode due to the reduced amount of gas being measured (<140 μg). Oxygen isotopic analyses are reported in standard δ notation, where δ 18  The glove box used for sample loading was a Plas-Labs continuous nitrogen flow model. This achieved low moisture and oxygen levels (<0.1 wt% O 2 ). Weighing the samples during loading proved problematic due to pressure fluctuations compromising the balance. A normalization procedure has therefore been adopted using the 0.5 mg value for particle C0068 obtained during initial sample preparation at the SPring-8 synchrotron facility before dispatch to the UK. C0068 was measured in its entirety as single measurement and gave a 17% yield (Table 1). A value of 17% is reasonable in view of the average 12.1% achieved on the measured CYs (Y-82162,82) and 25.2% measured on the CIs (Orgueil) ( Table 1).

Blank correction procedure
Blank correction data for all the samples analysed in this study are given in the Supplementary Information. The relatively small size of the Ryugu samples available for oxygen isotope analysis meant that it was necessary to apply a blank correction to all the samples analysed in this study 56 : where n T = total amount measured and is equal to n s + n b n b = amount of blank n s = amount of sample δ T = delta total amount δ b = delta blank δ s = delta sample The values of n b and δ b were determined by loading a Ryugu tray with only an obsidian standard present. The 'flexi' blanks were carried out as normal. A 5-min sample chamber blank was then run and the 4 μg of O 2 that was evolved during this procedure was run on the MAT 253 micro-volume. The results obtained were: δ 17 O = −5.15‰; δ 18 O = −9.95‰ and Δ 17 O = 0.02‰. As the amount of time that this blank was run for was greater than our usual lasering time of about 2 min, the blank correction applied was reduced to 2.4 μg of O 2 . Details of the blank correction applied to each analysis are given in the Supplementary Information.

δ 18 O variation in unhomogenized primitive CCs
In most cases, when determining the bulk oxygen isotope composition of a meteorite a relatively large chip of typically between 100 to 200 mg of the sample is crushed and homogenized. Aliquots of about 2 mg are then taken from this homogenized powder and analysed by laser fluorination. The aim is to determine a representative bulk composition for the meteorite. However, for some studies much smaller mg-sized fractions have been removed from primitive meteorites and analysed by laser fluorination. The range in δ 18 O obtained from these small subsamples often matches or exceeds the range observed in this study. Thus, in the case of NWA 7891 δ 18  In keeping with the results from the present investigation, the results from these meteorite studies indicat that where a sample displays a substantial level of inherent isotopic heterogeneity between mineral phases, mg-sized samples, without previous homogenization of a larger fraction of material, will show a large degree of δ 18 O heterogeneity.

δ 18 O variation related to modal variations in Ryugu particles
As discussed in the main text, Ryugu particles analysed in this study show a significant range in δ 18 O values (11.46 to 19.30‰). Calculations show that this variation can be fully explained in terms of the heterogeneous distribution of the main oxygen-bearing phases, phyllosilicate, magnetite and dolomite (Supplementary Information).

Calculations undertaken to model the influence of terrestrial contamination on the Δ 17 O composition of CI chondrites
As discussed in the main text, the small difference between the Δ 17 O compositions of the Ryugu particles and CI chondrites is most probably related to terrestrial contamination of the CI meteorites. Two distinct approaches have been used to model the contamination of the CIsl (Supplementary Information). Both sets of calculations give essentially identical results.   47,48 analyses are thought to be representative of the oxygen isotope composition of primordial water ice. The composition of the Ryugu particles can be explained in terms of relatively extensive closed system exchange between 16 O-poor fluids interpreted to have a composition similar to matrix materials from Acfer 094 (refs. 47,48 ) and 16 O-rich solids such as those commonly found in CC meteorites. This evidence appears to be at odds with the proposal that asteroid Ryugu is of cometary origin 44 (see main text for further discussion). Mixing lines are CCAM 13 and Y&R 66 , which is the Young and Russell Line.

Data availability
All of the data relevant to this publication are available in