Early fluid activity on Ryugu inferred by isotopic analyses of carbonates and magnetite

Samples from asteroid Ryugu returned by the Hayabusa2 mission contain evidence of extensive alteration by aqueous fluids and appear related to the CI chondrites. To understand the sources of the fluid and the timing of chemical reactions occurring during the alteration processes, we investigated the oxygen, carbon and 53Mn–53Cr systematics of carbonate and magnetite in two Ryugu particles. We find that the fluid was initially between 0 and 20 °C and enriched in 13C, 17O and 18O, and subsequently evolved towards lighter carbon and oxygen isotopic compositions as alteration proceeded. Carbonate ages show that this fluid–rock interaction took place within approximately the first 1.8 million years of Solar System history, requiring early accretion either in a planetesimal less than ∼20 km in diameter or within a larger body that was disrupted and reassembled. The aqueous activity responsible for carbonate formation on Ryugu happened much earlier—less than 1.8 million years after CAI formation—than estimates (4–6 Myr) from carbonaceous chondrite meteorites. Ryugu’s parent body either was smaller than ∼20 km in diameter or was disrupted before reaching the high temperatures required.

Samples from asteroid Ryugu returned by the Hayabusa2 mission contain evidence of extensive alteration by aqueous fluids and appear related to the CI chondrites. To understand the sources of the fluid and the timing of chemical reactions occurring during the alteration processes, we investigated the oxygen, carbon and 53 Mn- 53 Cr systematics of carbonate and magnetite in two Ryugu particles. We find that the fluid was initially between 0 and 20 °C and enriched in 13 C, 17 O and 18 O, and subsequently evolved towards lighter carbon and oxygen isotopic compositions as alteration proceeded. Carbonate ages show that this fluid-rock interaction took place within approximately the first 1.8 million years of Solar System history, requiring early accretion either in a planetesimal less than ∼20 km in diameter or within a larger body that was disrupted and reassembled.
The Hayabusa2 mission returned approximately 5.4 g of material from the C-type asteroid Ryugu. This material is highly aqueously altered and resembles the rare CI (Ivuna-type) chondrite meteorites, with abundant Mg-phyllosilicate, pyrrhotite, magnetite and carbonate signifying extensive fluid evolution on Ryugu's parent body [1][2][3][4][5] . Because aqueous alteration products such as magnetite and carbonate record information about the fluid from which they form, isotopic measurements of these components can be used to constrain the timing and characteristics of aqueous alteration of Ryugu materials.
In addition to their mineralogical similarities, the bulk oxygen isotopic compositions of the Ryugu particles and CI chondrites are also similar [3][4][5][6] . These values are primarily defined by the phyllosilicate Article https://doi.org/10.1038/s41550-022-01863-0 occurred in large (>50 km radius) parent bodies that accreted 3-4 Myr after CAI formation 11,12,36 . However, deriving initial 53 Mn/ 55 Mn ratios of carbonate based on in situ SIMS analyses requires standards that closely match the chemical composition of the target mineral to determine the Mn/Cr ratio accurately (that is, are 'matrix-matched'), particularly with regard to the Fe content of the carbonate [37][38][39] . Previous studies that targeted dolomite were performed using non-matrix-matched standards (primarily calcite) for the Mn/Cr ratio, which can affect the accuracy of the results 38,39 . In this work, we use matrix-matched calcite, dolomite and magnesite standards to obtain the 55 Mn/ 52 Cr ratios of respective mineral phases in Ryugu.
Ryugu particles A0037 and C0009, which were acquired from the first and second touchdown sites, respectively 3 , are dominated by minerals produced via aqueous alteration 1,3 . A0037 contains a much higher abundance of carbonate (21.2 vol%) than C0009 (1.8 vol%) 3 . Carbonates found in these two particles are primarily dolomite (CaMg(CO 3 ) 2 , Fig. 1a; see also

Oxygen isotopic composition of carbonate and magnetite
The oxygen isotopic compositions of dolomite and magnetite in particles A0037 and C0009 and Ca-carbonate in C0009 are shown in Fig. 2 and listed in Supplementary Tables 1 and 2. The oxygen isotopic compositions of dolomites mostly plot near the terrestrial fractionation line; however, several dolomite grains have positive Δ 17 O well resolved from 0‰, up to a maximum of +1.6 ± 0.3‰ (2σ) for a dolomite grain found in A0037 and +1.4 ± 0.9‰ (2σ) for a dolomite grain found in C0009. The δ 18 O values of dolomite grains are also somewhat variable, ranging from +25‰ to +34‰ in A0037 and from +22‰ to 27‰ in C0009. The range of oxygen isotopic compositions of Ryugu dolomite in A0037 and C0009 is in good agreement with earlier in situ analyses of CI chondrite dolomite 24 and Ryugu dolomite from other particles 4,5 (Fig. 2 25 .

Carbon isotopic compositions of carbonate
Dolomite in both Ryugu particles shows a wide range of δ 13 C values from 55.4‰ to 74.5‰ (Fig. 3 and Supplementary Table 3). Dolomite in A0037 appears to follow a bimodal distribution with δ 13 C peaks at ∼57 and ∼72‰. Dolomite (and some Ca-carbonate) in C0009 shows δ 13 C ranging from 64‰ to 75‰, with one Ca-carbonate enriched in δ 13 C at 97‰ ( Fig. 3 and Supplementary Table 3). These δ 13 C values are consistent with bulk measurements of Orgueil carbonates 33 and are similar to the compositions of carbonates in Tagish Lake 34 , but are more enriched in 13 10), and radiometric dating of secondary minerals has constrained the timing of fluid alteration to ∼4-6 Myr after Ca-Al-rich inclusion (CAI) formation 11,12 . However, the CI chondrites have been exposed to various degrees of terrestrial alteration, which appear to have affected the bulk oxygen isotopic compositions 6 . Ryugu particles therefore represent a unique opportunity to study pristine samples of hydrated asteroidal material.
Of the various alteration products found in hydrated extraterrestrial materials like returned Ryugu particles and CI chondrites, carbonate minerals are of particular interest because they can be dated using the short-lived 53 Mn- 53 Cr chronometer (half-life = 3.7 Myr), thereby tracking when liquid water was present and establishing a timescale for the accretion and alteration of carbonaceous planetesimals. Stable isotope studies of the major elements O and C can also provide insight into the sources of the fluids present, as well as the temperatures and reactions occurring in the asteroid or its progenitor. To preserve the petrological context and minimize consumption of precious Ryugu material, these analyses can be performed in situ with high spatial resolution using secondary ion mass spectrometry (SIMS) to sputter material from individual mineral grains with a spot size of ∼3-15 µm (Methods). This technique has also been applied to analyses of carbonate and other secondary minerals in CM and CI carbonaceous chondrites, which facilitates comparison between the returned Ryugu particles and previously studied meteorite samples.
The oxygen isotopic systematics of aqueous alteration products in carbonaceous chondrite meteorites have been extensively studied 10,13-25 and used to infer the extent of equilibration between co-accreted water ice, inferred to be 17,18 O-enriched 26  In addition, if two secondary phases with the same Δ 17 O are identified, the difference in δ 18 O between the two phases can be used to calculate an equilibrium formation temperature, based on the assumption that they precipitated from the same water composition 9,13,14,23 . Previously, CI chondrite formation temperatures have been estimated based upon the phyllosilicate-carbonate pair 9,10 . The oxygen isotopic compositions of magnetite, if found to be in equilibrium with other secondary phases, can be used in a similar fashion 23 .
The carbon isotopic compositions of carbonate have been used to infer the contributions of various C sources, such as insoluble and soluble organic matter 28,29 and isotopically heavy CO 2 -CO ices 30,31 , to the fluids in the carbonaceous chondrite parent bodies. In principle, carbon isotope compositions can also track reactions occurring within the fluid, such as methane formation and loss 19,22,23,32 , oxidation of organic material 18,20 and CH 4 -CO equilibration 23,33 . However, such studies have thus far been limited to carbonate from the CM (Mighei type), Tagish Lake (C2 ungrouped) 34 and Flensburg (C1 ungrouped) 25 chondrites; few in situ C isotopic measurements have been conducted on CI carbonate 35 .
The timing and duration of carbonate formation can be constrained in favourable circumstances by using Mn-Cr dating, and these ages can also be used to constrain the accretion time of the parent bodies from which samples originate. Carbonate minerals are an ideal target for this analysis as they strongly fractionate Mn from Cr during their formation, leading to large excesses in radiogenic 53 Cr through which a 53 Mn/ 55 Mn ratio at the time of carbonate formation can be inferred. Previous in situ studies of highly altered carbonaceous chondrites have found that most carbonate grains in these meteorite classes formed between 4 and 6 Myr after CAIs, leading to inferences, based upon models of planetesimal thermal evolution, that carbonate formation Article https://doi.org/10.1038/s41550-022-01863-0

Mn-Cr dating of carbonate
We measured 55 Mn/ 52 Cr and 53 Cr/ 52 Cr ratios for 20 spots on dolomite in A0037 and 16 spots on dolomite, breunnerite and Ca-carbonate in C0009 (Supplementary Table 4) and corrected for the relative sensitivity between Mn and Cr using matrix-matched, 52 Cr-implanted terrestrial carbonate standards 39 . The analysis conditions, standards development and Mn-Cr data on Ryugu carbonates are detailed in the Methods and Supplementary Text. The data show 53 Cr excesses that are well correlated with 55 Mn/ 52 Cr (Fig. 4), implying initial 53 Mn/ 55 Mn of (6.8 ± 0.5) × 10 -6 (MSWD = 0.7) for A0037 dolomite and (6.1 ± 0.9) × 10 -6 (MSWD = 0.3) for C0009 (all errors 2 standard errors). By calibrating these initial ratios relative to the initial 53 Mn/ 55 Mn ratio 41 of the D'Orbigny angrite, which has a well-defined absolute crystallization age 42    O-enriched fluid 26 was less equilibrated with 16 O-rich nebular solids 27 . As previously described 40 , Ca 2 is isolated in the matrix and surrounded by an iron sulfide rim (Fig. 1b), while other Ca-carbonates are found as chains and clusters of individual grains with no rims (Fig. 1c), further supporting that the formation conditions that produced Ca 2 were distinct from those that produced other Ca-carbonates 40 . We note that the petrology of Ca 2 closely resembles Type 1 calcites identified in the CM chondrites 20 , which have been interpreted as having formed in early stages of fluid alteration as pores produced by melting water ice were cemented by carbonate precipitation 17 . Figure 3 shows that Ca 2 is also enriched in 13 C at δ 13 C = +97‰, suggesting that carbon in the fluid was initially isotopically heavy and derived from outer Solar System CO 2 ices, similar to what has been inferred for some carbonaceous chondrites 30,31,34 . Therefore, we conclude that Ryugu accreted in the outer Solar System beyond the CO 2 ice line, consistent with previous observations of bulk H and N isotopes in Ryugu particles that suggest an outer Solar System origin 3 .
The population of Ca-carbonates in particle C0009 shows a range in Δ 17 O of ∼0 to +2.2‰, following a mass-independent trend that requires that the O isotopic composition of the fluid evolved over the course of Ca-carbonate precipitation. This is in contrast to calcite grains found in Orgueil 24 , which follow a mass-dependent trend (that is, have constant Δ 17 O) with a restricted range in δ 18 O. We suggest that this distinction reflects a difference in the extent of alteration processes experienced by Ryugu and by Orgueil: the Ca-carbonate in Ryugu recorded the progress of equilibration between fluid and 16 O-rich anhydrous silicate 27 , whereas calcite in Orgueil precipitated after this equilibration had been established.
Magnetite in both particles and the Ca 2 Ca-carbonate grain (Fig. 1a,b; see also Supplementary Fig. 6f in Yamaguchi et al. 40 ) in C0009 share the same Δ 17 O values (within uncertainty) that are higher than the Δ 17 O of dolomite and other Ca-carbonates, reflecting a less-equilibrated fluid composition. We conclude that magnetite and Ca-carbonates like Ca 2 were among the earliest minerals to precipitate during the alteration of the Ryugu protolith, predating most carbonate formation. If the Ca 2 Ca-carbonate and magnetite formed in equilibrium with the same fluid 40 , we estimate the formation temperature at this early stage of alteration using equilibrium thermometry of calcite and magnetite 46  We suggest the following order for the sequence of aqueous alteration on Ryugu: first, magnetite and Ca-carbonates like Ca 2 precipitated from aqueous fluids with high Δ 17 O at a temperature of <20 °C with the carbon isotopic composition of the fluid dominated by that of CO 2 ice. As the fluid continued to exchange oxygen with 16  Carbon and oxygen isotopic analyses performed on the same grains were used to explore correlations between the two isotopic systems. Figure 5 illustrates that δ 13 C is correlated with δ 18 O (upper panel) and Δ 17 O (lower panel), similar to trends observed for some CM chondrites 33 . This observation suggests that methane formation via serpentinization of the protolith followed by loss to space did not strongly affect the δ 13 C of Ryugu carbonate, as methane release would enrich 13 C in the fluid over time 20,32 . In contrast, we observe that carbonate formed from less-equilibrated water (for example, with higher δ 18 O and Δ 17 O) is also the most 13 C-enriched. One possible scenario could be that the initial unequilibrated fluid composition, presumably similar to the fluid recorded by Ca 2, evolved towards lower δ 13 C as the fluid interacted with and oxidized Ryugu's relatively 13 C-depleted organic matter 3 .
The old ages measured in Ryugu carbonate stand in contrast to ages obtained from carbonate in carbonaceous chondrites, most of which were thought to have formed 4-6 Myr after CAIs 11,12,36,45 . Ryugu carbonate in C0009 and A0037 is also slightly older than carbonate found in Flensburg 25 . This difference arises from our use of matrix-matched standards, as opposed to calcite standards used exclusively in previous studies, to determine the Mn/Cr of the carbonates. Had we corrected measured Mn + /Cr + using a relative sensitivity factor derived only from analyses of calcite, we would have obtained ages of 3.0 Myr and 3.5 Myr after CAI formation for A0037 and C0009 carbonate, respectively, approaching the range of ages previously determined for carbonates in carbonaceous chondrites 11,12,36,45 .  Article https://doi.org/10.1038/s41550-022-01863-0 These old carbonate formation ages suggest a substantially different formation scenario for Ryugu than those previously proposed for the asteroid parent bodies of carbonaceous chondrites. Our data show that aqueous fluids responsible for carbonate formation were active on Ryugu (or its progenitor asteroid) early in Solar System history, within the first ∼1.8 Myr after CAI formation. At that time, 26 Al in chondritic material was still at the level of 26 Al/ 27 Al ∼10 −5 , abundant enough to melt accreted ices and drive aqueous alteration. However, for 26 Al heating to not be so intensive as to cause water loss or even silicate melting and chemical differentiation, Ryugu must have initially accreted as a small asteroid that could effectively conduct heat away from its interior to cool itself by radiation. The inferred presence of co-accreted CO 2 ice constrains the initial temperature of the parent body to below the sublimation temperature of CO 2 . By modelling parent bodies accreting as mixtures of 50% chondritic material and 50% water ice 10,49 at an initial temperature of 78 K, we find that parent bodies accreting before 1.8 Myr must be smaller than 20 km in diameter for the internal temperature to remain below 400 K (refs. 50, 51). In such bodies, the interior 4 km reaches the melting point of water within 0.4 Myr after accretion, and remains warm enough to support liquid water for an additional 1.1-1.5 Myr.
Alternatively, it could be possible to form Ryugu components in a progenitor body larger than 20 km in diameter that was later disrupted by impact before reaching peak temperatures. Ryugu is an ∼1 km diameter asteroid, inferred, like many asteroids, to be a 'rubble pile', characterized by large internal void spaces and a low bulk density (1,190 ± 20 kg m −3 ) 52 . A multistage scenario of brecciation and reassembly is also supported by petrographic and shock characteristics observed in Ryugu particles 3,40,53 . This view is very different from prior estimates of parent body size and accretion times based upon younger carbonate ages, which suggested that CM and CI parent bodies were >50 km in diameter and accreted ∼3-3.5 Myr after CAI formation 11,12,36 .
An early formation scenario for C-type asteroids has implications for models seeking to understand the origins of the so-called isotopic dichotomy within the solar nebula. In this framework, the early Solar System was divided into two reservoirs: one characterized by isotopic compositions similar to those of the volatile-rich carbonaceous chondrites, and the other being isotopically similar to the compositions of volatile-depleted ordinary-chondrite, enstatite-chondrite and terrestrial materials 54 (collectively known as the non-carbonaceous isotopic reservoir). Whereas the non-carbonaceous group accreted from materials formed in the inner Solar System, the carbonaceous chondrite group is thought to have accreted in the outer Solar System, beyond the snow line. Based on 182 Hf-182 W ages of iron meteorites with carbonaceous chondrite affinities, it has been suggested that some planetesimals in the outer Solar System accreted within ∼1 Myr of CAI formation 55 . This timescale is consistent with such objects having melted and chemically differentiated into core-mantle structures due to 26 Al heating, and it is also consistent with the accretion time of NWA 011, a basaltic achondrite with carbonaceous chondrite affinities that accreted within 1.6 Myr of CAI formation 56 . Based on previous Mn-Cr dating of carbonates, it was thought that CM and CI chondrites escaped such heating by virtue of having accreted at later times, after most 26 Al had decayed. However, early formation for undifferentiated carbonaceous chondrite material, such as that from Ryugu, requires an explanation (for example, formation in a small body or early disruption by impact) for the simultaneous existence of differentiated and unmelted carbonaceous chondrite materials. Similarly, models of accretion and transport in the disc that invoke a late formation time for carbonaceous chondrite parent bodies 57 should consider the implications of early formation of these objects.

Petrographic characterization
The detailed scanning electron microscope (SEM) and electron probe microanalysis methods are reported by Yamaguchi et al. 40 . Laser micro-Raman spectroscopy 40 was used to attempt to distinguish whether Ca-carbonate in the Ryugu C0009 was calcite or aragonite, but not enough of the band peak was measured to distinguish between the CaCO 3 polymorphs. Article https://doi.org/10.1038/s41550-022-01863-0

Secondary ion mass spectrometry
In situ oxygen, carbon and Mn-Cr isotopes analyses of Ryugu carbonates and magnetite were performed using the UCLA CAMECA ims-1290 ion microprobe. The Ryugu A0037 and C0009 particles were mounted in epoxy and polished under dry conditions and coated with a thin layer of Au for SIMS analysis after petrographic characterization. After SIMS analysis, all pits were observed by SEM (Tescan Vega) at UCLA. Analyses found to overlap inclusions, cracks or voids were discarded. In all stable isotope analyses, calcite, magnetite and a suite of four dolomite standards of various Fe compositions were measured to quantify the instrumental mass fractionation as a function of Fe content 58,59 . The chemical and isotopic compositions of these reference materials are listed in Supplementary Table 5.
Oxygen isotope analysis. Oxygen isotope analyses were performed with a focused Cs + primary ion beam with 20 kV total accelerating voltage. Based on the size of the grains analysed, we used three different primary beam conditions: 3O-I (∼3 nA) with ∼15 µm spot for dolomite, 3O-II (∼700 pA) with ∼10 µm spot for dolomite and 3O-III (∼60 pA) with ∼3 µm spot for dolomite, Ca-carbonate and magnetite. A normalincidence electron gun was used for charge compensation. The oxygen isotopic compositions are reported as per mil deviations relative to standard mean ocean water (SMOW), which can be calculated by using The contribution of 16 OH − tail to the 17 O − signal was determined by using the ratio of the ion signal measured at the tail of the 16 OH − peak on the high mass end (mass unit = 17.00274 + 0.00361) to that measured at the centre of the 16 OH − peak and assuming a symmetric peak. This ratio was then multiplied by the 16 OH − count rate on the unknown samples recorded at the end of each spot analysis. All reported δ 17 O values have been corrected for the 16 OH − tail ( Supplementary Tables 1 and 2). The corrections for the 16 OH − tail range from approximately 0.1 to 1.0 ‰.
The compositional dependence of instrumental bias (that is, the 'matrix effect' on instrumental mass fractionation) was calibrated using an equation similar to that suggested in Śliwiński et al. 58 . Error bars represent 2σ analytical uncertainty, accounting for both the internal measurement precision (standard error of mean (SEM) over cycles measured) and the external reproducibility (SEM over standards measured) for bracketing measurements of the standards; σ 2 = (SEM unknown ) 2 + (SEM standard ) 2 .
Carbon isotope analysis. Carbon isotope analysis of carbonate was carried out using a focused Cs + ion primary beam of ∼600-700 pA. Secondary 12 C − and 13 C − ions were simultaneously detected using an FC and an EM, respectively. A normal-incidence electron gun was used for charge compensation. The typical count rate of 12 C − was ∼5.5-6 × 10 6 cps for MS1317J. The instrumental bias was corrected using MS1317J with a δ 13 C VPDB (Vienna PeeDee Belemnite) value of -1.20‰ ( 13 C/ 12 C = 0.011167) and optical calcite with a δ 13 C VPDB value of 1.42‰ ( 13 C/ 12 C = 0.011196) for dolomite and Ca-carbonate, respectively.
We defined the bias as where m and t stand for measured and true isotope ratios, respectively. Error bars represent 2σ analytical uncertainty, including both the internal measurement precision and the external reproducibility for standard measurements.

Mn-Cr isotope analysis.
Mn-Cr analyses of carbonates were carried out using a 1 nA 16 O 3 − primary ion beam generated by a Hyperion-II plasma ion source. For dolomite and magnesite with sufficient Mn content, secondary 52 Cr + , 53 Cr + and 55 Mn + ions were collected simultaneously Article https://doi.org/10.1038/s41550-022-01863-0 using two EMs (for 52 Cr + and 53 Cr + ) and an FC (for 55 Mn + ). A mass resolution power of ∼5,500 was used to separate 52 Cr + from 28 Si 24 Mg + and 53 Cr + from 52 CrH + . For dolomite and calcite with low Mn concentrations, 55 Mn + was collected using an EM in peak-switching mode. Analysis spots were presputtered using an 8 × 8 or a 4 × 4 µm raster to remove surface Cr contamination before focusing the beam to a tighter raster (5 × 5 or 2 × 2 µm) for the analysis, resulting in an effective spot size of ∼8 × 10 µm 2 . The instrumental mass fractionation for Cr was corrected by comparison to repeated measurements of the MS1317J dolomite, which contains trace amounts of terrestrial Cr ( 53 Cr/ 52 Cr = 0.113459) 60 . The relative sensitivity factor (RSF) between 55 Mn and 52 Cr is defined as RSF = ( 55 Mn/ 52 Cr) True ( 55 Mn/ 52 Cr) SIMS (2) where 'True' refers to the true ratio of concentrations in the mineral and 'SIMS' is the ratio of ion currents as measured during SIMS analysis. The RSF was determined using a combination of San Carlos Olivine and ion-implanted carbonate standards. Before the Mn-Cr analysis, the local distribution of 52 Cr was assessed using scanning ion imaging to avoid regions with high 52 Cr background, which can indicate contamination from Cr-rich phases. Calculation of the isochron slope was performed using the 'fit_bivariate' Python module, an implementation of the York et al. 61 line-fitting algorithm.

Data availability
Correspondence and requests for materials should be addressed to K.A.M. and N.M. All analytical data related to this manuscript will be put on the JAXA Data ARchives and Transmission System (https:// www.darts.isas.jaxa.jp/curation/hayabusa2) after a 1-year proprietary period.